Most recent entries from all

fkie_cve-2026-73558

vLLM is an inference and serving engine for large language models. Prior to 0.27.0, an integer overflow in blockIdx.x * 2 * d in activation_kernels.cu can cause act_and_mul_kernel to consume another batched user's input, allowing a request processed in the same inference batch to receive a partial or complete copy of another user's inference result. This issue is fixed in version 0.27.0.

GHSA-7m6h-x95x-82q5 — vLLM: Cross-User Data Leak Vulnerability

Affected: PyPI: vllm

### Summary An integer overflow in the act_and_mul_kernel kernel can cause the output of one user request to be incorporated into the response of another request within the same inference batch. Under certain conditions, the last request in a batch can receive a partial or complete copy of the first user's inference result, resulting in cross-user data leakage.

### Details The root cause is an integer overflow in the expression blockIdx.x * 2 * d at https://github.com/vllm-project/vllm/blob/ff712f6447093d07747c88680b9d006b119f5890/csrc/activation_kernels.cu#L82.

As a result, the computation for one user (User A) can incorrectly consume input data from another user (User B). In particular, when 2^32 is divisible by d, the overflow can cause User A's output to contain portions of User B's inference result. In some cases, User B's response may be copied entirely into User A's response.

This constitutes a severe cross-user information disclosure vulnerability and is straightforward to trigger. ### PoC We reproduced the issue using meta-llama/Llama-3.2-1B-Instruct, for which d = 8192.

Using the following configuration:

Batch size: 17 Sequence length: 16384

The final response in the batch becomes an exact copy of the first response in the batch, demonstrating complete cross-user data leakage.

### Impact This vulnerability enables cross-user information disclosure. An attacker can intentionally craft requests that are processed within the same inference batch as a victim's…

PYSEC-2026-3935 — vLLM: Cross-User Data Leak Vulnerability

Affected: PyPI: vllm

Using the following configuration:

Batch size: 17 Sequence length: 16384

The final response in the batch becomes an exact copy of the first response in the batch, demonstrating complete cross-user data leakage.

### Impact This vulnerability enables cross-user information disclosure. An attacker can intentionally craft requests that are processed within the same inference batch as a victim's…

RHSA-2026:82059 — Red Hat Security Advisory: Red Hat AI Inference 3.5.2 (ROCm)

libxml2: Stack Buffer Overflow in xmllint Interactive Shell Command Handling sqlite: SQLite: Arbitrary code execution via crafted FTS5 full-text search data sqlite: SQLite: Arbitrary code execution and crash via heap-based buffer overflow in FTS5 python: cpython: CPython: tarfile extraction filter bypass allows escaping the destination directory GDBusServer: glib2: GDBusServer pre-authentication DoS via unbounded SASL line buffering xdgmime: heap-based buffer overflow in _xdg_mime_magic_parse_magic_line() in xdgmimemagic.c urllib3: urllib3: Information disclosure via cross-origin redirects forwarding sensitive headers urllib3: urllib3: Denial of Service due to excessive HTTP response decompression vim: Vim: Arbitrary Code Execution via crafted directory names vim: Vim: Arbitrary code execution via crafted step-definition patterns vim: Vim: Arbitrary code execution via Python omni-completion acl: Symlink traversal privilege escalation via libacl functions linux-pam: Plaintext password recovery via timing discrepancy in pam_userdb module vim: Vim: Out-of-bounds Write in Spell File Word Count vim: Vim: Denial of Service via stack out-of-bounds write in spell_soundfold_sofo() vim: Vim: Arbitrary code execution via malicious docstrings in Python omni-completion glib: buffer over-read in glib/gvariant-serialiser.c via gvs_tuple_is_normal() glib: out-of-bounds read in glib/gdatetime.c:g_date_time_get_ymd via invalid GDateTime glib: buffer over-read in g_regex_replace() via glib/gre…