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-project vllm version v0.6.2 contains a vulnerability in the MessageQueue.dequeue() API function. The function uses pickle.loads to parse received sockets directly, leading to a remote code execution vulnerability. An attacker can exploit this by sending a malicious payload to the MessageQueue, causing the victim's machine to execute arbitrary code.

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.

This vulnerability allows remote attackers to execute arbitrary code on vulnerable installations of NetGain Systems Enterprise Manager v7.2.699 build 1001. Authentication is not required to exploit this vulnerability. The specific flaw exists within the handling of the content parameter provided to the script_test.jsp endpoint. A crafted content request parameter can trigger execution of a system call composed from a user-supplied string. An attacker can leverage this vulnerability to execute code under the context of the web service. Was ZDI-CAN-5080.

An OS Command Injection vulnerability was found in /landrecordsys/admin/dashboard.php in PHPGurukul Land Record System v1.0, which allows remote attackers to execute arbitrary code via the "Cookie" GET request parameter.

DLINK DIR-825 REVB 2.03 devices have an OS command injection vulnerability in the CGl interface apc_client_pin.cgi, which allows remote attackers to execute arbitrary commands via the parameter "wps_pin" passed to the apc_client_pin.cgi binary through a POST request.

Telesquare SDT-CW3B1 1.1.0 is affected by an OS command injection vulnerability that allows a remote attacker to execute OS commands without any authentication.

The web services of Softnext's products, Mail SQR Expert and Mail Archiving Expert do not properly validate user input, allowing unauthenticated remote attackers to inject arbitrary OS commands and execute them on the remote server.

The Comcast firmware on Cisco DPC3939 (firmware version dpc3939-P20-18-v303r20421746-170221a-CMCST) devices allows Network Processor (NP) Linux users to obtain root access to the Application Processor (AP) Linux system via shell metacharacters in commands.

Shell metacharacter injection vulnerability in /usr/www/include/ajax/GetTest.php in TerraMaster TOS before 3.0.34 leads to remote code execution as root.

tinyfiledialogs (aka tiny file dialogs) before 3.15.0 allows shell metacharacters (such as a backquote or a dollar sign) in titles, messages, and other input data. NOTE: this issue exists because of an incomplete fix for CVE-2020-36767, which only considered single and double quote characters.

Tenable Appliance 3.5 - 4.4.0, and possibly prior versions, contains a flaw in the simpleupload.py script in the Web UI. Through the manipulation of the tns_appliance_session_user parameter, a remote attacker can inject arbitrary commands.

totolink a3100r V5.9c.4577 is vulnerable to os command injection. The backend of a page is executing the "ping" command, and the input field does not adequately filter special symbols. This can lead to command injection attacks.

QNAP NAS application Proxy Server through version 1.2.0 allows remote attackers to run arbitrary OS commands against the system with root privileges.

The management interface for the Teltonika RUT9XX routers (aka LuCI) with firmware 00.03.265 and earlier allows remote attackers to execute arbitrary commands with root privileges via shell metacharacters in the username parameter in a login request.

A Remote Command Execution (RCE) vulnerability exists in HNAP1/control/SetNetworkTomographySettings.php of D-Link Router DIR-846 DIR846A1_FW100A43.bin and DIR846enFW100A53DLA-Retail.bin because backticks can be used for command injection when judging whether it is a reasonable domain name.

Tenda routers G1 and G3 v15.11.0.17(9502)_CN were discovered to contain a command injection vulnerability in the function formSetNetCheckTools. This vulnerability allows attackers to execute arbitrary commands via the hostName parameter.

QNAP QTS before 4.2.4 Build 20170313 allows attackers to gain administrator privileges and obtain sensitive information via unspecified vectors.

Tenda routers G1 and G3 v15.11.0.17(9502)_CN were discovered to contain a command injection vulnerability in the function formSetUSBShareInfo. This vulnerability allows attackers to execute arbitrary commands via the usbOrdinaryUserName parameter.

Accellion File Transfer Appliance version FTA_8_0_540 suffers from an instance of CWE-77: Improper Neutralization of Special Elements used in a Command ('Command Injection').

CyberPanel (aka Cyber Panel) before 2.3.5 allows Command Injection via completePath in the ProcessUtilities.outputExecutioner() sink. There is /filemanager/upload (aka File Manager upload) unauthenticated remote code execution via shell metacharacters.

getresetstatus in dns/views.py and ftp/views.py in CyberPanel (aka Cyber Panel) before 1c0c6cb allows remote attackers to bypass authentication and execute arbitrary commands via /dns/getresetstatus or /ftp/getresetstatus by bypassing secMiddleware (which is only for a POST request) and using shell metacharacters in the statusfile property, as exploited in the wild in October 2024 by PSAUX. Versions through 2.3.6 and (unpatched) 2.3.7 are affected.

In Sophos Web Appliance (SWA) before 4.3.1.2, a section of the machine's interface responsible for generating reports was vulnerable to remote command injection via functions, aka NSWA-1304.

An OS command injection vulnerability in Hardy Barth cPH2 eCharge Ladestation v1.87.0 and earlier, may allow an unauthenticated remote attacker to execute arbitrary commands on the system via a specifically crafted arguments passed to the connectivity check feature.

A command injection vulnerability exists in the mudler/localai version 2.14.0. The vulnerability arises from the application's handling of the backend parameter in the configuration file, which is used in the name of the initialized process. An attacker can exploit this vulnerability by manipulating the path of the vulnerable binary file specified in the backend parameter, allowing the execution of arbitrary code on the system. This issue is due to improper neutralization of special elements used in an OS command, leading to potential full control over the affected system.

ping.cgi on NETGEAR DGN2200 devices with firmware through 10.0.0.50 allows remote authenticated users to execute arbitrary OS commands via shell metacharacters in the ping_IPAddr field of an HTTP POST request.

QNAP QTS before 4.2.4 Build 20170313 allows attackers to gain administrator privileges and execute arbitrary commands via unspecified vectors.

IBM Security Verify Access Appliance 10.0.0 through 10.0.8 could allow a remote authenticated attacker to execute arbitrary commands on the system by sending a specially crafted request.

A CWE-306 "Missing Authentication for Critical Function" was discovered affecting the following devices manufactured by Advantech: EKI-6333AC-2G (<= 1.6.3), EKI-6333AC-2GD (<= v1.6.3) and EKI-6333AC-1GPO (<= v1.2.1). The vulnerability can be exploited by remote unauthenticated users capable of interacting with the default "edgserver" service enabled on the access point.

A CWE-78 "Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection')" was discovered affecting the following devices manufactured by Advantech: EKI-6333AC-2G (<= 1.6.3), EKI-6333AC-2GD (<= v1.6.3) and EKI-6333AC-1GPO (<= v1.2.1). The vulnerability can be exploited by remote unauthenticated users capable of interacting with the default "edgserver" service enabled on the access point and malicious commands are executed with root privileges. No authentication is enabled on the service and the source of the vulnerability resides in processing code associated to the "capture_packages" operation.

Dell Unity, version(s) 5.4 and prior, contain(s) an Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection') vulnerability. An unauthenticated attacker with remote access could potentially exploit this vulnerability, leading to Command execution.

An Improper Neutralization of Special Elements (in an OS command) issue was discovered in Geutebruck IP Camera G-Cam/EFD-2250 Version 1.11.0.12. An improper neutralization of special elements vulnerability has been identified. If special elements are not properly neutralized, an attacker can call multiple parameters that can allow access to the root level operating system which could allow remote code execution.

mySCADA myPRO: Versions 8.20.0 and prior has a feature where the firmware can be updated, which may allow an attacker to inject arbitrary operating system commands through a specific parameter.

Command Injection vulnerability in the web interface in McAfee Advanced Threat Defense (ATD) 3.10, 3.8, 3.6, 3.4 allows remote unauthenticated users / remote attackers to execute a command of their choice via a crafted HTTP request parameter.

OS Command Injection vulnerability in McAfee ePolicy Orchestrator (ePO) 5.9.0, 5.3.2, 5.3.1, 5.1.3, 5.1.2, 5.1.1, and 5.1.0 allows attackers to run arbitrary OS commands with limited privileges via not sanitizing the user input data before exporting it into a CSV format output.

An issue in ZIONCOM (Hong Kong) Technology Limited A7000R v.4.1cu.4154 allows an attacker to execute arbitrary code via the cig-bin/cstecgi.cgi to the settings/setPasswordCfg function.

In WhatsUp Gold versions released before 2023.1.3, a Remote Code Execution issue exists in Progress WhatsUp Gold. This vulnerability allows an unauthenticated attacker to achieve the RCE as a service account through NmApi.exe.

The Lenovo Service Framework Android application executes some system commands without proper sanitization of external input. In certain cases, this could lead to command injection which, in turn, could lead to remote code execution.

In Visual Tools DVR VX16 4.2.28.0, an unauthenticated attacker can achieve remote command execution via shell metacharacters in the cgi-bin/slogin/login.py User-Agent HTTP header.

An issue was discovered in Kaseya Unitrends Backup Appliance before 10.5.5. Multiple functions in the bpserverd daemon were vulnerable to arbitrary remote code execution as root. The vulnerability was caused by untrusted input (received by the server) being passed to system calls.

TP-Link Archer A7 Archer A7(US)_V5_210519 is affected by a command injection vulnerability in /usr/bin/tddp. The vulnerability is caused by the program taking part of the received data packet as part of the command. This will cause an attacker to execute arbitrary commands on the router.

A vulnerability was found in Brave UX for-the-badge and classified as critical. Affected by this issue is some unknown functionality of the file .github/workflows/combine-prs.yml. The manipulation leads to os command injection. The name of the patch is 55b5a234c0fab935df5fb08365bc8fe9c37cf46b. It is recommended to apply a patch to fix this issue. VDB-216842 is the identifier assigned to this vulnerability.

smalruby-editor v0.4.0 and earlier allows remote attackers to execute arbitrary OS commands via unspecified vectors.

TOTOLINK EX1200T V4.1.2cu.5215 is affected by a command injection vulnerability that can remotely execute arbitrary code.

TOTOLINK EX1200T V4.1.2cu.5215 contains a remote command injection vulnerability in function setDeviceMac of the file global.so which can control deviceName to attack.

TOTOLINK EX1200T V4.1.2cu.5215 contains a remote command injection vulnerability in function NTPSyncWithHost of the file system.so which can control hostTime to attack.

TOTOLINK EX1200T V4.1.2cu.5215 contains a remote command injection vulnerability in function setDeviceName of the file global.so which can control thedeviceName to attack.

The ZyXEL P660HN-T1A v1 TCLinux Fw $7.3.15.0 v001 / 3.40(ULM.0)b31 router distributed by TrueOnline has a command injection vulnerability in the Remote System Log forwarding function, which is accessible by an unauthenticated user. The vulnerability is in the ViewLog.asp page and can be exploited through the remote_host parameter.

A remote command execution (RCE) vulnerability was found in FeMiner wms V1.0 in /wms/src/system/datarec.php. The $_POST[r_name] is directly passed into the $mysqlstr and is executed by exec.