astral-tokio-tar is a tar archive reading/writing library for async Rust. In versions 0.5.6 and earlier, malformed PAX extensions were silently skipped when parsing tar archives. This silent skipping (rather than rejection) of invalid PAX extensions could be used as a building block for a parser differential, for example by silently skipping a malformed GNU “long link” extension so that a subsequent parser would misinterpret the extension. In practice, exploiting this behavior in astral-tokio-tar requires a secondary misbehaving tar parser, i.e. one that insufficiently validates malformed PAX extensions and interprets them rather than skipping or erroring on them. This vulnerability is considered low-severity as it requires a separate vulnerability against any unrelated tar parser. This issue has been fixed in version 0.6.0.

Rack is a modular Ruby web server interface. From versions 3.0.0.beta1 to before 3.1.21 and 3.2.0 to before 3.2.6, Rack::Utils.forwarded_values parses the RFC 7239 Forwarded header by splitting on semicolons before handling quoted-string values. Because quoted values may legally contain semicolons, a header can be interpreted by Rack as multiple Forwarded directives rather than as a single quoted for value. In deployments where an upstream proxy, WAF, or intermediary validates or preserves quoted Forwarded values differently, this discrepancy can allow an attacker to smuggle host, proto, for, or by parameters through a single header value. This issue has been patched in versions 3.1.21 and 3.2.6.

Squid is a caching proxy for the Web. Prior to version 7.5, due to premature release of resource during expected lifetime and heap Use-After-Free bugs, Squid is vulnerable to Denial of Service when handling ICP traffic. This problem allows a remote attacker to perform a reliable and repeatable Denial of Service attack against the Squid service using ICP protocol. This attack is limited to Squid deployments that explicitly enable ICP support (i.e. configure non-zero `icp_port`). This problem _cannot_ be mitigated by denying ICP queries using `icp_access` rules. This bug is fixed in Squid version 7.5.

telnetd in GNU inetutils through 2.7 allows an out-of-bounds write in the LINEMODE SLC (Set Local Characters) suboption handler because add_slc does not check whether the buffer is full.

libheif is a HEIF and AVIF file format decoder and encoder. Versions 1.21.2 and below contain a heap buffer overflow in MaskImageCodec::decode_mask_image(). When decoding a HEIF file containing a mask image (mski), the function copies the full iloc extent data into a pixel buffer using memcpy(dst, data.data(), data.size()). The copy length data.size() is determined by the iloc extent in the file (attacker-controlled), while the destination buffer is sized based on the declared image dimensions. Because no upper-bound check exists on the data length, a crafted file whose iloc extent exceeds the pixel buffer allocation overflows the heap. The vulnerable single-memcpy branch is reached when the mskC property specifies bits_per_pixel = 8 and the ispe property declares an even width ≥ 64 (so that stride == width), with no changes to default security limits or external codec plugins required. This issue has been fixed in version 1.22.0.

libheif is a HEIF and AVIF file format decoder and encoder. Versions 1.21.2 and prior contain a heap-buffer-overflow (write) vulnerability in the grid tile compositing, allowing an attacker to write 64 bytes of fully attacker-controlled data past the end of a chroma plane heap allocation by crafting a HEIF/AVIF file with a 1×4 grid of odd-height tiles. The overflow is triggered during normal image decoding with default build configuration. The written bytes are chroma (Cb/Cr) pixel values from the attacking tile, giving the attacker full control over the overflow content. This issue has been fixed in version 1.22.0.

libheif is a HEIF and AVIF file format decoder and encoder. In versions 1.21.2 and below, a crafted 800-byte HEIF sequence file causes an infinite loop in Box_stts::get_sample_duration(), consuming 100% CPU indefinitely with zero progress, leading to DoS. The loop has no iteration limit or timeout and is triggered during file open (parsing) - before any user interaction or image decoding. The process stays alive (no crash, no error logged), making it invisible to crash-based monitoring. This issue has been fixed in version 1.22.0.

libheif is a HEIF and AVIF file format decoder and encoder. In versions 1.21.2 and below, a crafted 792-byte HEIF sequence file with samples_per_chunk=0 in the stsc box causes an unsigned integer underflow in the Chunk constructor (m_last_sample = 0 + 0 - 1 = UINT32_MAX), mapping all samples to an empty chunk and resulting in a denial of service. When any sample is accessed, the library reads from index 0 of an empty std::vector, causing a guaranteed SEGV (null-page read). The file parses successfully without producing an error; the crash occurs on the first frame access. This issue has been fixed in version 1.22.0.

SciTokens C++ is a minimal library for creating and using SciTokens from C or C++. Prior to version 1.4.1, scitokens-cpp is vulnerable to an authorization bypass in path-based scope validation. The enforcer used a simple string-prefix comparison when checking whether a requested resource path was covered by a token's authorized scope path. Because the check did not require a path-segment boundary, a token scoped to one path could incorrectly authorize access to sibling paths that merely started with the same prefix. This issue has been patched in version 1.4.1.

SciTokens C++ is a minimal library for creating and using SciTokens from C or C++. Prior to version 1.4.1, scitokens-cpp is vulnerable to an authorization bypass when processing path-based scopes in tokens. The library normalizes the scope path from the token before authorization and collapses ".." path components instead of rejecting them. As a result, an attacker can use parent-directory traversal in the scope claim to broaden the effective authorization beyond the intended directory. This issue has been patched in version 1.4.1.

Memray is a memory profiler for Python. Prior to Memray 1.19.2, Memray rendered the command line of the tracked process directly into generated HTML reports without escaping. Because there was no escaping, attacker-controlled command line arguments were inserted as raw HTML into the generated report. This allowed JavaScript execution when a victim opened the generated report in a browser. Version 1.19.2 fixes the issue.

pydicom is a pure Python package for working with DICOM files. Versions 2.0.0-rc.1 through 3.0.1 are vulnerable to Path Traversal through a maliciously crafted DICOMDIR ReferencedFileID when it is set to a path outside the File-set root. pydicom resolves the path only to confirm that it exists, but does not verify that the resolved path remains under the File-set root. Subsequent public FileSet operations such as copy(), write(), and remove()+write(use_existing=True) use that unchecked path in file I/O operations. This allows arbitrary file read/copy and, in some flows, move/delete outside the File-set root. This issue has been fixed in version 3.0.2.

MariaDB server is a community developed fork of MySQL server. An authenticated user can crash MariaDB versions 11.4 before 11.4.10 and 11.8 before 11.8.6 via a bug in JSON_SCHEMA_VALID() function. Under certain conditions it might be possible to turn the crash into a remote code execution. These conditions require tight control over memory layout which is generally only attainable in a lab environment. This issue is fixed in MariaDB 11.4.10, MariaDB 11.8.6, and MariaDB 12.2.2.

Devise is an authentication solution for Rails based on Warden. Prior to version 5.0.3, a race condition in Devise's Confirmable module allows an attacker to confirm an email address they do not own. This affects any Devise application using the `reconfirmable` option (the default when using Confirmable with email changes). By sending two concurrent email change requests, an attacker can desynchronize the `confirmation_token` and `unconfirmed_email` fields. The confirmation token is sent to an email the attacker controls, but the `unconfirmed_email` in the database points to a victim's email address. When the attacker uses the token, the victim's email is confirmed on the attacker's account. This is patched in Devise v5.0.3. Users should upgrade as soon as possible. As a workaround, applications can override a specific method from Devise models to force `unconfirmed_email` to be persisted when unchanged. Note that Mongoid does not seem to respect that `will_change!` should force the...

In Juju from version 3.0.0 through 3.6.18, when a secret owner grants permissions to a secret to a grantee, the secret owner relies exclusively on a predictable XID of the secret to verify ownership. This allows a malicious grantee which can request secrets to predict past secrets granted by the same secret owner to different grantees, allowing them to use the resources granted by those past secrets. Successful exploitation relies on a very specific configuration, specific data semantic, and the administrator having the need to deploy at least two different applications, one of them controlled by the attacker.

In Juju from version 3.0.0 through 3.6.18, the authorization of the "secret-set" tool is not performed correctly, which allows a grantee to update the secret content, and can lead to reading or updating other secrets. When the "secret-set" tool logs an error in an exploitation attempt, the secret is still updated contrary to expectations, and the new value is visible to both the owner and the grantee.

An authorization bypass vulnerability in the Vault secrets back-end implementation of Juju versions 3.1.6 through 3.6.18 allows an authenticated unit agent to perform unauthorized updates to secret revisions. With sufficient information, an attacker can poison any existing secret revision within the scope of that Vault secret back-end.

A race condition in the secrets management subsystem of Juju versions 3.0.0 through 3.6.18 allows an authenticated unit agent to claim ownership of a newly initialized secret. Between generating a Juju Secret ID and creating the secret's first revision, an attacker authenticated as another unit agent can claim ownership of a known secret. This leads to the attacking unit being able to read the content of the initial secret revision.

In NLnet Labs Unbound 1.22.0 up to and including 1.25.1, when downstream DNS-over-QUIC (DoQ) is enabled, the first two bidirectional streams on a new QUIC connection (stream_id 0 and 4) bypass the per-stream 'quic-size' gate entirely, and large input buffers are allocated later, after only the 2-byte length prefix has been received from the initial streams. As a result, a remote client can make Unbound exceed the configured 'quic-size' limit with low-cost input. Using only one connection and two streams, each sending a declared 65535-byte length prefix and then holding the streams open, a client can already trivially make Unbound roughly allocate double that amount. This is a remote availability issue / memory-accounting bypass in the downstream DoQ implementation that leads to denial of service for new DoQ clients. This vulnerability needs Unbound to be compiled with DoQ support ('--with-libngtcp2') and the 'quic-port' to be configured for the listening interfaces.

NGINX Open Source and NGINX Plus have a vulnerability in the ngx_http_mp4_module module, which might allow an attacker to trigger a buffer over-read or over-write to the NGINX worker memory resulting in its termination or possibly code execution, using a specially crafted MP4 file. This issue affects NGINX Open Source and NGINX Plus if it is built with the ngx_http_mp4_module module and the mp4 directive is used in the configuration file. Additionally, the attack is possible only if an attacker can trigger the processing of a specially crafted MP4 file with the ngx_http_mp4_module module. Note: Software versions which have reached End of Technical Support (EoTS) are not evaluated.