| CVE |
Vendors |
Products |
Updated |
CVSS v3.1 |
| In the Linux kernel, the following vulnerability has been resolved:
drm/amdgpu: disallow multiple FENCE chunks in one submit
amdgpu_cs_pass1() dispatches on chunk_id once per chunk without
rejecting repeated ids. p->uf_bo is a single-slot field, so a
submission carrying two AMDGPU_CHUNK_ID_FENCE chunks runs
amdgpu_cs_p1_user_fence() twice, and the second run overwrites
p->uf_bo with a freshly referenced BO without dropping the reference
taken by the first.
amdgpu_cs_parser_fini() only unrefs the final p->uf_bo, so every FENCE
chunk but the last leaks a BO reference. The leaked BO outlives handle
close and process exit.
Reject duplicate FENCE chunks the same way commit fec5f8e8c6bc
("drm/amdgpu: disallow multiple BO_HANDLES chunks in one submit") did
for p->bo_list.
(cherry picked from commit 665b1fc2a1845206408f9a2c6da67101789edb82) |
| The Breeze Cache WordPress plugin before 2.5.13 does not sanitise a value taken from the request before using it to build the paths of the files it caches, allowing unauthenticated attackers to create files at arbitrary locations on the server, outside the intended cache directory. |
| GitLab has remediated a vulnerability in the GitLab AI Gateway component affecting all versions of the AI Gateway from 18.9.0 to 19.0.12, 19.1 to 19.1.7, and 19.2 to 19.2.2 that could have allowed an authenticated user with Duo Agent Platform access to redirect model requests to an externally-controlled endpoint via crafted model metadata, resulting in the disclosure of Google Vertex AI or AWS Bedrock cloud service credentials. |
| In Bouncy Castle for Java before 1.85, OpenPGP CFB quick-check oracle active on symmetric/session-key paths. This issue also affects Bouncy Castle for Java LTS before 2.73.12, and Bouncy Castle for Java FIPS (BC-FJA) before bcpg-fips 1.0.13 (1.0.X series), 2.0.13 (2.0.X series) and 2.1.13 (2.1.X series). |
| Unrestricted Upload of File with Dangerous Type in the product photo upload in Roskus Prospero Flow CRM before 5.16.0 allows an authenticated user holding the create product permission (routine Seller role) to execute arbitrary JavaScript in the application origin. The photo validation rule classifies the file only by its content (magic bytes) and rejects only a fixed list of PHP extensions, while ProductSaveController::save() names the stored file using the client-supplied extension and copies it into the public web root. A file that begins with an image header and carries an HTML extension passes validation, is stored under public/asset/upload/product/, and is served with a text/html content type, turning the upload into first-party stored script execution. |
| A vulnerability was found in MAA-AI MaaMCP up to 1.1.1.dev6+g2e4a41287. The affected element is the function save_pipeline/load_pipeline of the file pipeline_tools.py. Performing a manipulation results in path traversal. The attack can be initiated remotely. The exploit has been made public and could be used. The patch is named c93ef45cba75295eba26d9ff1ffb9202a91c6150. To fix this issue, it is recommended to deploy a patch. |
| In Bouncy Castle for Java before 1.85, CMS verifySignatures returns true for SignedData with zero signers. This issue also affects Bouncy Castle for Java LTS before 2.73.12, and Bouncy Castle for Java FIPS (BC-FJA) before bcpkix-fips 1.0.12 (1.0.X series), 2.0.12 (2.0.X series) and 2.1.12 (2.1.X series). |
| A flaw has been found in RooCodeInc Roo-Code up to 3.51.1. This issue affects the function path.resolve of the file src/core/tools/ApplyPatchTool.ts of the component ApplyPatchTool. This manipulation causes path traversal. It is possible to initiate the attack remotely. The exploit has been published and may be used. Multiple isses were reported to the vendor beforehand. They explain, that "they all apply to Roo Code, a project we no longer support - the repository was archived a while ago, and we don't encourage anyone to use it." This vulnerability only affects products that are no longer supported by the maintainer. |
| The User Frontend WordPress plugin before 4.3.10 does not restrict access to its user directory search endpoint, allowing unauthenticated attackers to retrieve the email address and phone number of every registered user, including administrators. |
| The Ultimate Member WordPress plugin before 2.13.0 does not validate a submitted role selection when it cannot resolve the set of roles a profile form permits, and screens the value against the site's registered role names rather than against the form's own allow-list, allowing unauthenticated users who register through the Ultimate Member WordPress plugin before 2.13.0's own form to grant themselves arbitrary capabilities and reach administrator-equivalent access. |
| cpp-httplib is a C++ header-only HTTP/HTTPS library. In version 0.49.0, the chunked-response trailer output path writes trailer header names and values directly to the socket without validating them, allowing CRLF sequences in a trailer field to inject additional headers or split the HTTP response. Unlike every other header-writing path in the library, the trailer-writing code applies none of the field-name and field-value checks that reject carriage return and line feed, so an application that places attacker-influenced data into a chunked response trailer emits attacker-controlled CRLF onto the wire. This enables HTTP response splitting, letting an attacker forge response headers or inject a second response. This issue is fixed in version 0.50.0. |
| The tar extraction routines in moby/go-archive (Unpack, UnpackLayer, Untar/UntarUncompressed, and the ApplyLayer helpers) do not confine filesystem operations to the destination directory. The extractor decides where each archive entry lands using lexical string checks and then performs the filesystem operation on a path that is resolved by the OS, so links introduced by the archive can be followed out of the destination directory. An attacker who controls the contents of an archive can create or overwrite files at arbitrary paths writable by the extracting process. |
| In SiSDK v2026.6.0 and earlier, high network traffic loads can cause a dropped ACK leading to a denial of service. This is only present for EFR32MG24 and EFR32MG26 devices running concurrent multiprotocol Zigbee and Thread. |
| Observable discrepancy vulnerability in Seres Software syWEB allows Account Footprinting.
This issue affects syWEB: through 27082026.
NOTE: The vendor was contacted and it was learned that the product is not supported. |
| Improper neutralization of input during web page generation ('cross-site scripting') vulnerability in Seres Software syWEB allows Reflected XSS.
This issue affects syWEB: through 27082026.
NOTE: The vendor was contacted and it was learned that the product is not supported. |
| Affected versions of Flowintel incorrectly attempted to validate login email addresses by calling Email(email). That does not perform WTForms field validation; it merely constructs a validator object.
Consequently, malformed attacker-controlled email input could continue through the login process and be written to security-relevant logs. The vulnerable code inserted the supplied email into both a warning log and the custom audit logger. Since CR/LF characters were not escaped, an unauthenticated attacker could potentially inject additional physical log lines or forge misleading log entries.
The patch corrects the validation call to Email()(form, form.email), changes the standard logging call to parameterized logging, and introduces _sanitize_log_fragment() so carriage returns and line feeds are encoded instead of creating new records.
Version impacted >=3.3.0 |
| Affected versions of Flowintel construct timeline HTML using attacker-controllable MISP object fields such as:
*
object UUID;
*
object name;
*
attribute value;
*
attribute type;
*
comment;
*
first/last seen values;
*
IDS flag.
Those values were concatenated directly into HTML strings before rendering. The upstream commit explicitly states that DOMPurify removed XSS vectors but still allowed other HTML elements, such as forms, through.
The fix replaces direct string interpolation with DOM construction via document.createElement() and assigns all attacker-controlled values using textContent. The headline is similarly converted to escaped HTML through a temporary element.
Version impacted =>3.3.0 |
| Affected versions of Flowintel expose the /my_assignment/user API endpoint to any authenticated API user. The endpoint accepts a user_id parameter identifying the user whose assignments should be returned, but previously had no role restriction beyond general API authentication.
As a result, a lower-privileged authenticated user could potentially query another user’s assignment information by supplying that user’s identifier.
The fix changes:
method_decorators = [api_required]
to:
method_decorators = [admin_or_org_admin_required, api_required]
so only administrators or organization administrators can perform cross-user assignment queries.
Version impacted =>3.3.0 |
| Affected versions of Flowintel contain an authorization flaw in the administrative user-edit API.
The existing authorization check correctly prevented an organization administrator from editing users in another organization, but it did not prevent them from editing a full administrator within their own organization. As a result, an org admin could modify that full administrator account, including changing its password. The upstream commit explicitly describes the issue as:
“Org admin can change the password of a full admin in the same organization.”
The fix adds a higher-privilege boundary check:
if user_to_edit.is_admin(): return ... 403
so organization administrators can no longer modify full administrator accounts.
Version impacted >=3.3.0 |
| Affected versions of Flowintel contain an insecure direct object reference / broken object-level authorization issue across numerous task endpoints.
The routes generally received both a case identifier and a task identifier, but previously they did not enforce that the task actually belonged to the supplied case. As a result, an authenticated user with editor-level access to one case could potentially substitute the ID of a task from another case and invoke operations against that foreign task.
The patch introduces task_case_bound_required, which loads both objects and returns 404 unless the task belongs to the requested case. This protection is applied to edit, delete, note, assignment, status, file, export, MISP-linking, subtask, external-reference, and other task-related endpoints.
The fix also adds explicit checks that a requested note_id belongs to the current task before returning or exporting it, closing related cross-object access paths.
Version impacted =>3.3.0 |
