vLLM versions prior to 0.32.0 contain a vulnerability where an out-of-vocabulary token ID equal to the vocabulary size can bypass input validation and cause a CUDA device-side assert, crashing the engine. This affects models whose vocabulary size is an exact multiple of the embedding padding size (64), such as Llama-2-family models with a vocab size of 32000. The issue occurs specifically when using a single GPU (tensor parallel size 1) configuration. The flaw was fixed in vLLM version 0.32.0 by correcting the maximum accepted token ID validation.
A vulnerability in filelock versions 3.22.0 through 4.2.0 allows a local user with write access to a lock marker file to forge the process ID (PID) recorded in the marker. This causes another process, running as the lock owner or root, to mistakenly consider the lock stale and remove the live marker, resulting in two processes concurrently holding the lock. The issue arises when the marker file is group- or other-writable due to explicit mode settings or permissive umask values in sticky or non-removable directories. The vulnerability has a medium severity with a CVSS score of 4.7 and impacts integrity but not confidentiality or availability. A fix is available in version 4.2.1.
CVE-2026-118649 is a prototype pollution vulnerability in n8n versions before 2.43.5 and 2.44.x before 2.44.1. It allows an authenticated user with workflow creation and execution privileges to manipulate a client-supplied workflow ID, such as using '__proto__', to bypass project ownership checks. This enables the user to list, read, modify, and delete Data Table rows across all projects in the instance.
n8n versions prior to 1.123.85, 2.43.5, and 2.44.1 contain an authorization bypass vulnerability. Authenticated users can exploit inline sub-workflow nodes with non-boolean disabled values to access credentials without proper permission. This flaw allows attackers who know a credential ID to run Execute Workflow nodes in inline JSON mode, bypassing permission checks and exfiltrating decrypted credentials to attacker-controlled servers.
CVE-2026-118247 is a prototype pollution vulnerability in Highcharts JS, a JavaScript charting library. The flaw exists in DataTable single-cell writers that use untrusted column IDs without filtering out __proto__ or constructor keys. This allows writes to Object.prototype or the Object constructor, potentially altering application control flow or disrupting rendering. The vulnerability affects versions prior to 13.1.0, Dashboards before 4.2.3, and Grid Pro before 3.2.0. It requires control over the table or component column ID and a cell, point, or API write to exploit. The issue is fixed in Highcharts 13.1.0, Dashboards 4.2.3, and Grid Pro 3.2.0.
CVE-2026-79656 is a vulnerability in Grafana OSS where sensitive OAuth and OIDC tokens are logged unredacted under certain conditions. This occurs when access or ID tokens are invalid JWTs, missing (at debug level), or not strings, exposing tokens in server logs. The exposure is limited to the scopes granted to the OAuth client. The issue affects multiple versions of Grafana OSS and has been fixed in versions 12.4.11, 13.0.9, 13.1.6, and 13.2.2.
CVE-2026-103406 is a critical OS command injection vulnerability in the AutomatedProvisioner component of Juniper Networks Session Smart Conductor. It allows a remote attacker to execute arbitrary commands as root by sending a specially crafted salt minion ID. The vulnerability affects multiple versions prior to fixed releases. The flaw is identified as CWE-78 and has a CVSS v3.1 score of 9.8, indicating high impact on confidentiality, integrity, and availability.
A vulnerability in identrail prior to version 1.0.2 allows an authenticated tenant to bypass authorization checks on the GitHub installation ID parameter. This flaw permits reading private repository data and metadata of other organizations if identrail's GitHub App is installed there. The issue arises because the installation ID is accepted without verifying ownership by the caller's workspace. The vulnerability is fixed in version 1.0.2.
A vulnerability in the Linux kernel's ksmbd component involves the FS_POSIX_INFORMATION structure where the FileSysIdentifier field is not properly assigned in smb2_get_info_filesystem(). This can cause residual data from previous responses to be leaked to clients during compound SMB2 requests. The issue arises because the response buffer is reused without clearing the FileSysIdentifier field, potentially exposing unintended data. The fix involves correctly reporting the file system ID as returned by statfs().
A vulnerability in certain API endpoints allowed creation of order payments referencing gift cards by using an unchecked auto-incremented internal ID. This flaw enabled attackers to reference gift cards from different organizers and obtain their secret. Although the gift card itself was not charged, triggering a refund on the payment could increase the gift card's value.
A vulnerability in the Linux kernel's XFS filesystem capability check allowed quota bypass by changing a file's group ID to another group the user belonged to, even when quotas were enforced. The issue was caused by using a capability check function with different semantics than intended, allowing quota accounting checks to be bypassed in some cases such as via NFS clients. The flaw has been fixed by replacing the capability check with the correct function to prevent quota evasion and unnecessary SELinux audit messages.
OAuthLib contains a vulnerability in its implementation of the OAuth 2.0 device authorization grant (RFC 8628). Specifically, the token endpoint does not validate the device_code parameter, allowing an attacker who knows a valid client ID for a public client enabled for the device workflow to obtain a bearer token without a valid device_code. This affects versions of vdirsyncer from 0.19.3_3 onward.
CVE-2026-99212 is a high-severity vulnerability in the Linux kernel's RDS (Reliable Datagram Sockets) RDMA connection management. It involves improper handling of RDMA connection IDs on route failure, where an ID that RDS still owns can be prematurely destroyed, leading to use-after-free conditions. This flaw was fixed by ensuring the connection ID remains owned by RDS until proper shutdown, preventing premature destruction.
A vulnerability in the Linux kernel futex subsystem related to FUT_OFF_MMSHARED futexes has been fixed. The issue allowed processes to wait on shared futexes referencing a different memory management (MM) context, causing waiters on a freed MM to consume wakeups intended for a newly allocated MM at the same address. The fix involves keying FUT_OFF_MMSHARED futexes with a unique 64-bit per-MM ID to prevent this confusion. This vulnerability affects Linux kernel version 5.7.
A vulnerability in the Linux kernel's ring-buffer implementation caused an off-by-one error in the sub-buffer array base calculation. This flaw led to an incorrect index being returned by rb_meta_subbuf_idx(), resulting in a shifted sub-buffer ID array with a NULL entry that could be dereferenced when userspace maps trace_pipe_raw. The issue was fixed by correcting the index calculation to ensure consistency between mapping functions.