vLLM is an inference and serving engine for large language models (LLMs). Prior to version 0.14.1, a Server-Side Request Forgery (SSRF) vulnerability exists in the `MediaConnector` class within the vLLM project's multimodal feature set. The load_from_url and load_from_url_async methods obtain and process media from URLs provided by users, using different Python parsing libraries when restricting the target host. These two parsing libraries have different interpretations of backslashes, which allows the host name restriction to be bypassed. This allows an attacker to coerce the vLLM server into making arbitrary requests to internal network resources. This vulnerability is particularly critical in containerized environments like `llm-d`, where a compromised vLLM pod could be used to scan the internal network, interact with other pods, and potentially cause denial of service or access sensitive data. For example, an attacker could make the vLLM pod send malicious requests to an internal `llm-d` management endpoint, leading to system instability by falsely reporting metrics like the KV cache state. Version 0.14.1 contains a patch for the issue.

PyTorch is a Python package that provides tensor computation. Prior to version 2.10.0, a vulnerability in PyTorch's `weights_only` unpickler allows an attacker to craft a malicious checkpoint file (`.pth`) that, when loaded with `torch.load(..., weights_only=True)`, can corrupt memory and potentially lead to arbitrary code execution. Version 2.10.0 fixes the issue.

Knockpy 4.1.1 contains a CSV injection vulnerability that allows attackers to inject malicious formulas into CSV reports through unfiltered server headers. Attackers can manipulate server response headers to include spreadsheet formulas that will execute when the CSV is opened in spreadsheet applications.

A flaw was found in Glib's content type parsing logic. This buffer underflow vulnerability occurs because the length of a header line is stored in a signed integer, which can lead to integer wraparound for very large inputs. This results in pointer underflow and out-of-bounds memory access. Exploitation requires a local user to install or process a specially crafted treemagic file, which can lead to local denial of service or application instability.

In GnuPG before 2.5.17, a stack-based buffer overflow exists in tpm2daemon during handling of the PKDECRYPT command for TPM-backed RSA and ECC keys.

Mitigation bypass in the Privacy: Anti-Tracking component. This vulnerability affects Firefox < 147.0.2.

Suricata is a network IDS, IPS and NSM engine. Prior to versions 8.0.3 and 7.0.14, specially crafted traffic can cause Suricata to consume large amounts of memory while parsing DNP3 traffic. This can lead to the process slowing down and running out of memory, potentially leading to it getting killed by the OOM killer. Versions 8.0.3 or 7.0.14 contain a patch. As a workaround, disable the DNP3 parser in the suricata yaml (disabled by default).

A flaw was found in kubevirt. A user within a virtual machine (VM), if the guest agent is active, can exploit this by causing the agent to report an excessive number of network interfaces. This action can overwhelm the system's ability to store VM configuration updates, effectively blocking changes to the Virtual Machine Instance (VMI). This allows the VM user to restrict the VM administrator's ability to manage the VM, leading to a denial of service for administrative operations.

pnpm is a package manager. Prior to version 10.28.2, when pnpm processes a package's `directories.bin` field, it uses `path.join()` without validating the result stays within the package root. A malicious npm package can specify `"directories": {"bin": "../../../../tmp"}` to escape the package directory, causing pnpm to chmod 755 files at arbitrary locations. This issue only affects Unix/Linux/macOS. Windows is not affected (`fixBin` gated by `EXECUTABLE_SHEBANG_SUPPORTED`). Version 10.28.2 contains a patch.

A heap-based buffer overflow vulnerability was found in ImageMagick in versions prior to 7.0.11-14 in ReadTIFFImage() in coders/tiff.c. This issue is due to an incorrect setting of the pixel array size, which can lead to a crash and segmentation fault.