CVE Vulnerability Catalog

LightLLM through version 1.2.0 contains a memory exhaustion vulnerability in the NCCL control channel when started with --pd_trans_mode nccl. Unauthenticated attackers can call the exposed_set_value method to store unbounded key-value pairs without size limits, causing the worker process to crash and triggering node failure.

LightLLM through 1.2.0 multimodal deployments expose an unauthenticated RPyC cache service with pickle deserialization enabled on all interfaces. Attackers can send crafted serialized objects to exposed cache methods to execute arbitrary code with service privileges.

LightLLM through 1.2.0 contains a remote code execution vulnerability in the router profiler service when started with --enable_profiling flag. The service exposes an unauthenticated RPyC server with pickle deserialization enabled, allowing attackers to execute arbitrary code by sending crafted serialized objects to the profiler command queue.

A vulnerability was detected in Ziroom ZHOME A0101 1.0.1.0. This affects the function set_syslog of the file /api/ZRnetwork/set_syslog. The manipulation of the argument conloglevel/log_size results in command injection. The attack may be performed from remote. The exploit is now public and may be used. The vendor was contacted early about this disclosure but did not respond in any way.

The device's update mechanism includes conditions that allow unauthorized software packages to be accepted as authentic. During boot, the stock done function disables signature verification in the OPKG configuration before restoring optional packages from a writable, unsigned feed. Additionally, the publicly distributed SDK contains the production private key whose corresponding public key is trusted by both stable and beta firmware builds.

The device's update mechanism retrieves metadata for software updates over an unencrypted HTTP connection and stores portions of that metadata for later use. A management interface subsequently returns this stored value in a JSON response, and the web interface responsible for displaying update information inserts that value directly into the page as HTML, allowing attacker-controlled metadata to be interpreted as script content. Additionally, the same authenticated origin provides an interface capable of executing system-level commands with root privileges.

U-Boot before 2026.10-rc5 contains an out-of-bounds memory access in dhcp6_parse_options() that fails to validate SERVERID and CLIENTID option lengths from DHCPv6 packets. Attackers on the local network can send crafted DHCPv6 ADVERTISE or REPLY packets during netboot to corrupt memory and crash the bootloader.

U-Boot before 2026.10-rc5 contains a use-after-free vulnerability in the httpc_recv_cb() function within the lwIP wget implementation. When HTTP data storage fails, the callback frees the connection PCB but returns ERR_BUF instead of ERR_ABRT, causing the TCP input path to access released memory and crash the bootloader.

U-Boot before 2026.10-rc5 contains a buffer overflow in the nfs_readlink_reply() function in net/nfs-common.c when processing NFS server responses. A malicious NFS server can send crafted READLINK replies with negative or oversized symlink length values to corrupt memory and crash the bootloader.

U-Boot before 2026.10-rc5 contains a buffer overflow in the nfs_read_reply() function in net/nfs-common.c that allows attackers to corrupt memory by supplying crafted NFS READ reply lengths. A malicious NFS server can exploit signed integer handling to bypass length validation and write far past the destination buffer, crashing the bootloader or corrupting memory.

The "supplier_no" parameter used in the business allocation search feature is vulnerable to time-based blind SQL injection.

The "reportType" parameter in the product summary report feature within the balancing reports section is susceptible to a time-based blind SQL injection vulnerability.

U-Boot before 2026.10-rc3 contains an out-of-bounds write vulnerability in read_slotted_partition() that fails to validate image size against partition bounds. An attacker with physical access can supply crafted boot media with oversized headers to write past the load buffer into bootloader memory on devices without Android Verified Boot protection.

U-Boot before 2026.10-rc4 contains an integer overflow vulnerability in sqfs_read_directory_table() when allocating the directory table buffer. An attacker can supply a crafted SquashFS image with an attacker-controlled superblock metablks_count value that causes heap buffer under-allocation and out-of-bounds writes, corrupting heap memory and crashing the bootloader.

U-Boot through 2026.10-rc5 contains an out-of-bounds write vulnerability in the video_display_rle8_bitmap function in drivers/video/video_bmp.c. An attacker can supply a crafted RLE8-compressed BMP image to corrupt memory adjacent to the framebuffer and crash the bootloader.

U-Boot before 2026.10-rc3 with CONFIG_IP_DEFRAG enabled contains an out-of-bounds write vulnerability in the __net_defragment() function in net/net.c. Remote attackers can send a crafted IP fragment with non-zero offset and More-Fragments flag set during netboot to corrupt adjacent memory and crash the bootloader.

The file export endpoint allows any unauthenticated attacker to export arbitrary database tables by sending a crafted POST request.

The application accepts user-supplied session identifiers and does not regenerate the session ID after authentication. This allows an attacker to predefine a session ID and reuse it after victim authentication, resulting in session takeover.

An attacker can construct a request that, if issued by another application user, will cause JavaScript code supplied by the attacker to execute within the user's browser in the context of that user's session with the application. This is a classic cross-site scripting (XSS) vulnerability.

The TMS file upload endpoint fails to enforce server-side file type restrictions, allowing an attacker to upload and execute arbitrary PHP files on the web server.