A vulnerability in Open Neural Network Exchange (ONNX) versions from 1.2.1 up to 1.22.0 allows a denial of service (DoS) via a crash during shape inference. The issue occurs when ConvTranspose shape inference processes a weight tensor with fewer than three dimensions without a kernel_shape attribute, leading to an invalid indexing operation. This flaw affects the convTransposeShapeInference_opset11() function and the current inference path. The vulnerability can be triggered by an untrusted model that passes the onnx.checker.check_model() validation but causes a crash in onnx.shape_inference.infer_shapes(). The issue is fixed in version 1.22.0.
A vulnerability in Open Neural Network Exchange (ONNX) prior to version 1.22.0 allows an application crash due to improper shape inference processing of malformed RNN, LSTM, and GRU nodes. The issue arises because the function RNNShapeInference_opset1_to_6() does not verify that the first input has rank three, leading to out-of-bounds access on protobuf shape data. This can cause termination of applications processing untrusted models that pass model checks.
Python Social Auth versions prior to 6.2.0 contain a vulnerability in the OpenIdConnectAuth backend where the OpenID Connect nonce is stored globally rather than being bound to the browser session and OAuth state. This allows an attacker who obtains a victim's unused authorization code to reuse it by initiating a separate login with a valid state, bypassing nonce validation and potentially logging in as the victim if PKCE is disabled. The issue is fixed in version 6.2.0.
A denial of service vulnerability exists in the rust-zip library versions from 1.3.0 up to but not including 9.0.0. The ZipArchive::extract function improperly uses an attacker-controlled ZIP64 uncompressed_size value to allocate memory, which can cause the process to abort before validating the entry. This abort cannot be caught or contained, allowing repeated denial of service attacks. The issue is fixed in version 9.0.0.
A vulnerability in the Rust zip library versions from 7.2.0 until 9.0.0 allows an attacker to exploit symbolic link validation flaws during ZIP extraction. The issue arises because the library uses lexical prefix comparison instead of resolving intermediate symlinks, enabling crafted symlink chains to escape the intended extraction directory. On Windows, additional privilege conditions can allow single-link escapes. This flaw can lead to attacker-controlled files being created or overwritten with the extracting process's privileges. The vulnerability is fixed in version 9.0.0.
The Rust zip library versions from 5.1.0 until 9.0.0, when built with the legacy-zip feature, have a vulnerability where output capacity can be reserved based on an attacker-controlled ZIP64 uncompressed size before validating the compressed stream. This can cause a panic due to capacity overflow in certain decoder paths, leading to unhandled panics that terminate archive-processing services. The issue is fixed in version 9.0.0.
The Rust zip library versions from 0.6.0 up to but not including 9.0.0 contain a vulnerability in the AES entry reader where HMAC verification is improperly performed. Specifically, the HMAC is only checked when an attacker-controlled data counter reaches zero, allowing certain conditions to bypass authentication checks. This can cause applications using password-based decryption methods to accept modified plaintext without detecting authentication errors. The issue is fixed in version 9.0.0.
The Rust zip library (rust-zip) through version 9.0.4 has a vulnerability where ZipArchive treats Central Directory File Header filenames as authoritative, but ZipStreamReader extracts files using Local File Header filenames without reconciling the differences. This discrepancy allows a crafted ZIP archive to appear benign during inspection but cause the extractor to write files to attacker-chosen paths, potentially overwriting critical files. This issue is fixed in version 9.0.5.
Netty netty-tcnative versions before 2.0.84.Final contain a Server Name Indication (SNI) hostname spoofing vulnerability. This flaw allows remote unauthenticated clients to manipulate hostnames via JNI NewStringUTF decoding by sending overlong encodings in the ClientHello server_name field. The vulnerability can cause confusion in SNI-based certificate selection, access control decisions, and logging, potentially leading to incorrect hostname handling.
A denial of service vulnerability exists in netty-tcnative versions before 2.0.84.Final. The flaw is in the tcn_write_to_bytebuffer BIO write callback, which overreports bytes consumed. This can be triggered by remote TLS 1.3 peers sending queued post-handshake records such as NewSessionTicket or KeyUpdate, causing truncated flights, connection failures, or JVM process aborts.
A use-after-free vulnerability exists in netty-tcnative versions before 2.0.84.Final when built against OpenSSL 3.x. The flaw occurs during TLS session ticket key rotation, where remote clients performing TLS handshakes can trigger OpenSSL to read freed HMAC keys, potentially causing crashes or corruption in ticket processing.
Netty netty-tcnative versions before 2.0.84.Final have a heap buffer overflow vulnerability in the tcn_write_to_bytebuffer BIO callback. This flaw arises from improper handling of the non-application ring buffer write index, allowing remote TLS peers to trigger an out-of-bounds memory copy. Exploitation can cause the JVM process to crash during handshake, renegotiation, or post-handshake message processing.
OpenJPEG versions 2.4.0 through 2.5.4, and git master through commit 8314119b, contain an out-of-bounds read and write vulnerability in the function opj_dwt_encode_and_deinterleave_v() located in src/lib/openjp2/dwt.c. This vulnerability can be exploited by an attacker to cause a denial of service condition.
Mattermost versions 11.7.0 through 11.7.10 contain an authorization flaw where the API does not properly restrict which built-in roles a team admin can assign to team members. This allows a team admin to grant permissions beyond the configured permission scheme via the update-team-member-roles REST API endpoint. The vulnerability has a low severity rating with limited impact on confidentiality and integrity.
A vulnerability in Brotli was discovered where it incorrectly handled certain compressed streams. This flaw could allow a remote attacker to cause a denial of service or execute arbitrary code by providing specially crafted input. The issue affects multiple Ubuntu package versions and has been addressed in updated releases. Users are advised to update to the fixed package versions to mitigate the risk.
Mattermost versions 11.7.x through 11.10.x (up to specified patch levels) contain a vulnerability in the WebSocket msgpack decoder where allocation limits are not properly enforced. This flaw allows an authenticated remote attacker to cause excessive server memory allocation by sending a crafted binary WebSocket frame with a large declared Bin32 payload length. The issue leads to a denial-of-service condition due to resource exhaustion.
Mattermost versions 11.7.x up to 11.7.10 contain a vulnerability where reserved system post types are not properly validated during scheduled post creation and publication. This flaw allows an authenticated channel member to publish messages that appear as trusted system events via the scheduled post endpoint. The vulnerability has a low severity rating and does not impact confidentiality or availability.
Mattermost versions up to 11.9, including 11.1.9, 11.5.8, and 11.10.7.0, contain a vulnerability where the Zoom chat line length is not properly validated before slicing. This flaw allows an authenticated attacker to crash the Agents plugin by summarizing a Zoom chat transcription with short continuation lines.
A code injection vulnerability exists in AllenAI AllenNLP up to version 2.10.1 in the Registrable.resolve_class_name function. This flaw allows remote attackers to manipulate the Type argument, potentially leading to code execution. The vulnerability affects unsupported versions, and no vendor response or patch is available. Public exploit code exists, increasing the risk of exploitation.