Operating systems stack vulnerabilities · Page 47

Weak PBKDF2 parameters for password-encrypted private-key export in jose

Use of Password Hash With Insufficient Computational Effort in the jose library allows an attacker who obtains a password-encrypted private-key export to recover the private key through an offline password-cracking attack. When jose exports a password-encrypted private key with jose_jwk:to_der/2 (JOSE.JWK.to_der/2), the jose_jwk_der:to_binary/3 function in src/jwk/jose_jwk_der.erl serializes the key as a PKCS#8 EncryptedPrivateKeyInfo using the PBES2 scheme with hardcoded key-derivation parameters: PBKDF2-HMAC-SHA256 with an iteration count of 2048 and an 8-byte (64-bit) random salt. Neither value can be overridden by the caller. Both value

• Update to 1.11.13

Infinite recursion (CPU denial-of-service) in JOSE.JWK.shared_secret/2 on mismatched key types

jose_jwk:shared_secret/2 (exposed in Elixir as JOSE.JWK.shared_secret/2) derives an ECDH shared secret from two JWKs. Its first function clause only matches when both JWKs carry the same key type (kty) module, and its fallback clause re-invokes itself as shared_secret(from(YourJWK), from(MyJWK)). Because from/1 returns an already-parsed jose_jwk record unchanged, passing two JWKs whose key types differ (for example an EC key and an RSA key) makes the fallback clause re-enter with identical arguments, producing an unbounded tail-recursive loop that never returns and never crashes. An application that derives a shared secret from an externally

• Update to 1.11.13

Decompression bomb via unbounded zlib:inflate in jose_jwe_zip:uncompress/2

Improper Handling of Highly Compressed Data (Data Amplification) vulnerability in potatosalad jose (erlang-jose) allows a remote attacker to exhaust a node's memory with a small crafted JWE token. The jose_jwe_zip:uncompress/2 function in src/jwe/jose_jwe_zip.erl decompresses the DEFLATE-compressed plaintext of a JWE that carries the optional zip: DEF header by calling zlib:inflate/2 on the entire attacker-controlled stream, with no bound on the output size or the inflation ratio. Raw DEFLATE reaches ratios around 1000:1 on highly repetitive input, so a few-kilobyte token inflates to hundreds of megabytes or gigabytes in a single call. Deco

• Update to 1.11.13

Asterisk: Out-of-Bounds Read in Q.931 Information Element Parser (H.323 Addon)

Asterisk is an open source private branch exchange and telephony toolkit. Prior to Asterisk 20.20.1, 21.12.3, 22.10.1, 23.4.1 and Certified Asterisk 20.7-cert11, 22.8-cert3, insufficient bounds checking in the `chan_ooh323` `Q.931` `information-element parser` permits an `out-of-bounds read` that may allow an unauthenticated attacker to crash Asterisk using carefully crafted `setup packets`. The `chan_ooh323` addon channel driver must be explicitly compiled and installed, which is not the default, and its port must be publicly reachable. This issue is fixed in Asterisk 20.20.1, 21.12.3, 22.10.1, 23.4.1 and Certified Asterisk 20.7-cert11, 22.8

• Update to 1:22.10.1+dfsg+~cs6.17.60671434-1

Bleichenbacher padding oracle in RSAES-PKCS1-v1_5 decryption in jose

The pure-Erlang RSAES-PKCS1-v1_5 decryption in the jose_jwa_pkcs1 module is a Bleichenbacher padding oracle. eme_pkcs1_decode/2 returns the recovered message when a decrypted block has valid PKCS#1 v1.5 padding (a leading 0x00 0x02, a padding string of at least eight non-zero bytes, and a 0x00 separator) and returns a distinct error result otherwise, and rsaes_pkcs1_decrypt/2 additionally returns {error, {badsize, _}} when the length check fails. The implementation applies no implicit-rejection countermeasure (RFC 8017 §7.2.2): it does not substitute a deterministic pseudo-random plaintext on a padding failure, so a one-bit padding-conformanc

• Update to 1.11.13

Attacker-controlled AES-GCM authentication tag length allows tag forgery in jose

The pure-Erlang AES-GCM fallback in jose_jwa_aes derives the number of verified authentication tag bits from the attacker-supplied tag itself. In the {aes_gcm, Bits} clause of block_decrypt/4, the incoming tag is guarded only by is_binary/1, with no check on its length. Verification in gcm_block_decrypt/6 then computes TBits = bit_size(T) and compares only those bits of the real 128-bit tag. An attacker who controls a JWE token can truncate the authentication tag to as little as 8 bits, so only 8 bits are ever checked. This reduces forgery resistance from 2^-128 to 2^-8, and a forged ciphertext is reliably accepted as authentic within 256 att

• Update to 1.11.13

Asterisk: LDAP Filter Injection in res_config_ldap via SIP Username (Unauthenticated Information Disclosure)

Asterisk is an open source private branch exchange and telephony toolkit. Prior to Asterisk 20.20.1, 21.12.3, 22.10.1, 23.4.1 and Certified Asterisk 20.7-cert11, 22.8-cert3, the LDAP realtime driver `res_config_ldap` may permit `LDAP filter injection` through a `SIP username`, allowing an unauthenticated attacker to make arbitrary LDAP queries using carefully crafted SIP packets and disclose information. The `res_config_ldap` module must be loaded and configured to access an LDAP server, which is not the default. This issue is fixed in Asterisk 20.20.1, 21.12.3, 22.10.1, 23.4.1 and Certified Asterisk 20.7-cert11, 22.8-cert3.

• Update to 1:22.10.1+dfsg+~cs6.17.60671434-1

Tornado: Event loop blocking and excess memory use from chunked bodies with many small chunks

Tornado is a Python web framework and asynchronous networking library. Prior to 6.6.0, Tornado decodes HTTP/1.1 chunked transfer encoding by processing body data that has already arrived without yielding to the event loop. A remote unauthenticated peer can send millions of one-byte chunks to monopolize the process event loop and stall every other connection served by that process, while lowering max_body_size only limits the duration of each request. For handlers that do not use @stream_request_body, retaining each chunk as a separate object can consume roughly twenty times the received body size, and SimpleAsyncHTTPClient has the same event-

• Update to 6.6

Asterisk: Heap-use-after-free in Asterisk PJSIP TCP/SDP handling when TCP connection closes during SDP processing

Asterisk is an open source private branch exchange and telephony toolkit. Prior to Asterisk 20.20.1, 21.12.3, 22.10.1, 23.4.1 and Certified Asterisk 20.7-cert11, 22.8-cert3, an authenticated attacker may cause a heap-use-after-free in `PJSIP TCP/SDP` handling by sending a `SIP INVITE` over a connection-oriented transport and disconnecting before Asterisk responds with the `200 OK`. So far, the crash has only been reproduced when Asterisk is built with `Address Sanitizer`, which detects the issue. `Address Sanitizer` is a development tool with significant performance overhead and, according to the maintainer, is not used in production systems.

• Update to 1:22.10.1+dfsg+~cs6.17.60671434-1

Asterisk: Reflected XSS in Phone Provisioning HTTP Error Pages

Asterisk is an open source private branch exchange and telephony toolkit. Prior to Asterisk 20.20.1, 21.12.3, 22.10.1, 23.4.1 and Certified Asterisk 20.7-cert11, 22.8-cert3, an unauthenticated attacker may use reflected cross-site scripting in `phone provisioning HTTP error pages` to retrieve an authorized user's `AMI session cookie` and impersonate that user when `res_phoneprov` is enabled. The Asterisk HTTP web server, Asterisk Manager Interface (AMI), and AMI access via HTTP must all be enabled; each is disabled by default. The attacker must be able to connect to the HTTP server, which binds only to `127.0.0.1` by default even when enabled

• Update to 1:22.10.1+dfsg+~cs6.17.60671434-1

ggml-org llama.cpp Mirostat Sampler llama-sampler.cpp assertion

A security flaw has been discovered in ggml-org llama.cpp up to 0.7.0. Affected by this vulnerability is an unknown functionality of the file src/llama-sampler.cpp of the component Mirostat Sampler. The manipulation results in reachable assertion. It is possible to launch the attack remotely. The project was informed of the problem early through a pull request but has not reacted yet.

Consul vulnerable to an authorization bypass in health check registration

HashiCorp Consul and Consul Enterprise are vulnerable to an authorization bypass in health check registration that may allow an authenticated user to modify health checks belonging to another service on the same node. This vulnerability (CVE-2026-115804) is fixed in Consul Community Edition 2.1.0 and Consul Enterprise 2.1.0 and 1.21.19.

• Update to 1.21.19
• Update to 2.1.0

Consul vulnerable to a denial of service in HTTP APIs

HashiCorp Consul and Consul Enterprise are vulnerable to a denial of service in the HTTP API that may allow an unauthenticated attacker to exhaust memory on a Consul agent. This vulnerability (CVE-2026-105626) is fixed in Consul Community Edition 2.1.0 and Consul Enterprise 2.1.0 and 1.21.19.

• Update to 1.21.19
• Update to 2.1.0

Consul vulnerable to an authorization bypass in service mesh registration

HashiCorp Consul and Consul Enterprise are vulnerable to an authorization bypass in service mesh proxy registration. This may allow a user with service write permission to apply Envoy escape-hatch configuration without the required mesh write permission. This vulnerability (CVE-2026-105608) is fixed in Consul Community Edition 2.1.0 and Consul Enterprise 2.1.0 and 1.21.19.

• Update to 2.1.0

Consul vulnerable to an authorization bypass in the agent's service mesh leaf certificate API

HashiCorp Consul and Consul Enterprise 1.17.0 up to 2.0.4 are vulnerable to an authorization bypass in the agent's service mesh leaf certificate API that may allow an attacker without a valid ACL token to retrieve a cached service leaf certificate and its private key. This vulnerability (CVE-2026-105765) is fixed in Consul Community Edition 2.1.0 and Consul Enterprise 2.1.0 and 1.21.19.

• Update to 1.21.19
• Update to 2.1.0

ONNX: Denial of service via unchecked expansion of packed sub-byte tensors in ONNX Python APIs

Open Neural Network Exchange (ONNX) is an open standard for machine learning interoperability. Prior to 1.22.0, _unpack_2bit() and _unpack_4bit() resize unpacked arrays according to attacker-controlled TensorProto.dims without verifying that raw_data or int32_data encodes the declared number of elements. Materializing an untrusted UINT2, INT2, UINT4, INT4, or FLOAT4E2M1 tensor through onnx.numpy_helper.to_array() or onnx.reference.ReferenceEvaluator can therefore expand a small packed payload into a disproportionately large allocation and exhaust memory. This issue is fixed in version 1.22.0.

• Update to 1.22.0

ONNX: Null Pointer Dereference in ScanInferenceFunction Due to Missing Required Attribute

Open Neural Network Exchange (ONNX) is an open standard for machine learning interoperability. From 1.4.0 until 1.22.0, ScanInferenceFunction, ScanInferenceFunction_opset8, and ScanInferenceFunction_opset9 do not safely validate the required num_scan_inputs attribute. A missing attribute is dereferenced as a null pointer, while a negative value is converted to size_t and can reach an exceptional oversized allocation path. An untrusted model can therefore terminate or disrupt shape inference, including a negative-value model that can pass onnx.checker.check_model(). This issue is fixed in version 1.22.0.

• Update to 1.22.0

ONNX: Out-of-bounds read in ConvTranspose shape inference via undersized weight rank

Open Neural Network Exchange (ONNX) is an open standard for machine learning interoperability. From 1.2.1 until 1.22.0, ConvTranspose shape inference processes a weight tensor with fewer than three dimensions without a kernel_shape attribute by deriving an empty effective_kernel_shape and indexing it for the input's spatial dimensions. The flaw affects convTransposeShapeInference_opset11() and the current inference path, and an untrusted model that can pass onnx.checker.check_model() can crash onnx.shape_inference.infer_shapes(), causing denial of service without demonstrated confidentiality or integrity impact. This issue is fixed in version

• Update to 1.22.0

ONNX: Out-of-Bounds Read in RNN/GRU/LSTM Shape Inference (Old Opsets, Missing Rank Check)

Open Neural Network Exchange (ONNX) is an open standard for machine learning interoperability. Prior to 1.22.0, RNNShapeInference_opset1_to_6() processes malformed RNN and LSTM nodes from opsets 1 through 6 and GRU nodes from opsets 3 through 6 without verifying that the first input has rank three. The function accesses dim(0) and dim(1) on an undersized protobuf shape, which can read outside the intended repeated field and terminate an application when onnx.shape_inference.infer_shapes() processes an untrusted model that can pass onnx.checker.check_model(). This issue is fixed in version 1.22.0.

• Update to 1.22.0

CVE-2026-116612

A flaw was found in GStreamer Editing Services (gst-editing-services). A GES project file can embed an arbitrary GStreamer pipeline description as an effect definition. When such a file is loaded, either explicitly by an editing application or automatically by a media player or file-manager thumbnailing service that uses GStreamer autoplugging, the pipeline description is parsed and instantiated without restriction on which elements may be used or which properties may be set. A remote attacker who can deliver a crafted project file to a victim's system can trigger reading or writing of arbitrary local files, issue outbound network requests to

• Update to 1.28.8
• Update to 1.28.8-1