The kernel-rt packages provide the Real Time Linux Kernel, which enables fine-tuning for systems with extremely high determinism requirements.
Security Fix(es):
* kernel: ipv4: raw: reject IP_HDRINCL packets with ihl < 5 (CVE-2026-64114)
* kernel: fuse: re-lock request before returning from fuse_ref_folio() (CVE-2026-64266)
* kernel: ipv4: fib: free fib_alias with kfree_rcu() on insert error path (CVE-2026-64572)
* kernel: ALSA: seq: close a re-opened queue timer in the destructor (CVE-2026-68202)
* kernel: xfs: bounds-check buffer log item's dirty bitmap (CVE-2026-80536)
* kernel: inet: frags: strip GSO state from fragments before reassembly (CVE-2026-80590)
* kernel: vfio/pci: clear vdev->msi_perm after freeing it on init failure (CVE-2026-89777)
For more details about the security issue(s), including the impact, a CVSS score, acknowledgments, and other related information, refer to the CVE page(s) listed in the References section.
Two critical remote code execution vulnerabilities in PHP, known as PHP2Shell (CVE-2026-109618 and CVE-2026-109619), were disclosed on October 21, 2026. CVE-2026-109618 involves corruption during out-of-memory handling that disrupts PHP while internal request data is partially initialized, causing invalid data processing. CVE-2026-109619 exploits a stale global state where request-owned internal state persists after memory is freed. Both vulnerabilities affect ARM64 systems and x86 systems without ASLR. Fixed versions include PHP 8.5.12, 8.4.27, 8.3.36, and 8.2.35. Over 2.6 million internet-facing PHP hosts were identified as vulnerable, with 42% running end-of-life branches lacking official patches. Early exploitation attempts have been observed, including scanning and webshell deployment targeting the upload-progress vulnerability path.
CVE-2026-112381 is a vulnerability in Apache Traffic Server that causes a crash due to a null pointer dereference. This occurs when a client POST request is half-closed while a chunked origin response is being reparsed in a reentrant write-ready callback. The issue is timing-dependent and was introduced by backports related to CVE-2021-32564. Affected versions include 9.2.15, 10.1.4, and 10.2.0. Fixed versions are 9.2.16, 10.1.5, and 10.2.1.
CVE-2025-55702 is a high-severity HTTP request/response smuggling vulnerability in Apache Traffic Server. It occurs when a client sends a POST request with the Expect: 100-continue header and the origin server responds before reading the request body. This causes Apache Traffic Server to misinterpret the unread body as a new request, enabling an attacker behind a shared front-end proxy to smuggle requests past the front end and desynchronize responses for other users. The vulnerability affects multiple versions of Apache Traffic Server from 8.0.0 through 8.1.11, 9.0.0 through 9.2.15, 10.0.0 through 10.1.4, and version 10.2.0. Fixed versions include 9.2.16, 10.1.5, and 10.2.1.
CVE-2026-78119 is a vulnerability in the Zephyr project's Nuvoton NPCX eSPI driver affecting versions 3.6.0 up to but not including 4.5.0. The flaw involves improper bounds checking when parsing Target-Attached-Flash (TAF) requests, allowing out-of-bounds writes and reads in the eSPI controller memory. This can lead to controller crashes, hangs, or corruption of driver state that controls flash protection. The vulnerability requires host firmware or bus-level control to exploit and is not reachable by ordinary host software. A fix is available in Zephyr version 4.5.0 that adds proper bounds checking and rejects invalid requests. No in-tree board enables the vulnerable feature by default.
CVE-2026-78118 is a race condition vulnerability in the Zephyr kernel's wait queue handling on SMP (multi-CPU) systems. The kernel helper z_waitq_head() reads the head of a thread wait queue without holding the necessary scheduler spinlock, leading to potential use-after-free and out-of-bounds writes. This can cause scheduler state corruption or kernel crashes. Exploitation requires local code execution, multiple CPUs, and precise timing, making crashes more common than controlled corruption. The issue is fixed in Zephyr version 4.5.0.
CVE-2026-78116 is a use-after-free vulnerability in the Broadcom AFBR-S50 time-of-flight sensor driver within Zephyr OS versions 4.3.0 up to but not including 4.5.0. The flaw occurs when sensor data evaluation fails, causing the same RTIO submission to be completed twice, leading to memory corruption and potential loss of availability. The issue can be triggered by environmental sensor conditions without requiring software or physical access. A fix is available in Zephyr version 4.5.0.
CVE-2026-105937 is a missing authorization vulnerability in Apache Pulsar's shadow topic feature. It allows an authenticated user with tenant admin role in their own tenant to create or update shadow topics bound to source topics in other tenants without authorization on those source topics. This can lead to unauthorized reading of retained messages and unintended deletion of source topic ledger data. The issue affects Apache Pulsar versions from 3.0.0 before 4.0.14 and from 4.1.0 before 4.2.5. Fixed versions disable shadow topics by default and require superuser role for cross-tenant shadow topic binding.
CVE-2026-105927 is an authorization vulnerability in Apache Pulsar's WebSocket service that allows an authenticated user with consume permission on a topic to inject messages into arbitrary topics via the dead letter policy configuration. The WebSocket service does not verify produce permissions on the dead letter topic, enabling unauthorized message injection. This vulnerability affects Apache Pulsar versions from 2.3.0 before 4.0.14 and from 4.1.0 before 4.2.5. The WebSocket service is disabled by default and can be disabled as a mitigation if not required. The vulnerability does not allow reading, modifying, or deleting existing messages, but injected messages may impact downstream applications depending on their processing logic.
Apache Pulsar versions from 2.8.1 before 4.0.14 and from 4.1.0 before 4.2.5 contain a critical vulnerability in the MultiRolesTokenAuthorizationProvider. When the anonymousUserRole fallback is enabled, the provider fails to verify the cryptographic signature of JSON Web Tokens (JWTs), allowing unauthenticated clients to forge or use unsigned tokens to gain arbitrary roles, including super-user privileges. This enables unauthorized read and modification of tenants, namespaces, topics, permissions, and broker configuration. Deployments using the default authorization provider or without the anonymousUserRole setting are not affected. The issue is fixed in versions 4.0.14 and 4.2.5 and later.
MikroTik RouterOS versions up to 7.23.6 and 7.24.5 are affected by a session token replay vulnerability in the WebFig interface. The flaw allows attackers to bypass authentication by replaying captured session tokens across new ECDH sessions due to insufficient session expiration and token validation. The vulnerability arises because session tokens are generated with fixed salts and validated without verifying channel context, enabling persistent administrative access until tokens are cleared. This issue extends the impact of a prior transport-layer exposure vulnerability (CVE-2025-61481) which remains unpatched.
Two critical remote code execution vulnerabilities, collectively known as PHP2Shell, affect PHP versions back to 5.4 and impact widely used web applications such as WordPress, Drupal, and Joomla. These vulnerabilities involve memory management and error handling components of PHP and allow unauthenticated attackers to execute arbitrary code. Exploitation attempts have been detected in the wild shortly after patches were released. The flaws also enable reading sensitive process memory, potentially exposing credentials and secrets. Patches are available in PHP versions 8.5.12, 8.4.27, 8.3.36, 8.2.35, and backports for 7.4, 8.0, and 8.1. The PHP Foundation and security firms urge immediate patching due to the severity and active exploitation attempts.
A heap use-after-free in the gf_odf_ac4_cfg_clean_list function (/odf/descriptors.c) of GPAC Project/MP4Box before 2.5 allows attackers to cause a Denial of Service (DoS) via supplying a crafted AC4 file.
A reachable exit vulnerability exists in NTPsec ntpd versions 1.1.4 through 1.2.5, allowing a remote attacker to cause a denial of service by crashing the daemon. This vulnerability impacts availability but does not affect confidentiality or integrity. The CVSS score is 7.5, indicating a high severity level.
CVE-2026-76481 is a high-severity vulnerability in Cisco Emergency Responder involving improper handling of exceptional conditions. This flaw is part of a group of issues identified through Cisco's internal security review and relates to CWE-703. The vulnerability affects multiple specific versions of Cisco Emergency Responder. It has a CVSS 3.1 score of 7.5, indicating high impact primarily on availability. No public exploits or known active exploitation have been reported.
CVE-2026-113800 is an origin validation error in the Electron framework affecting versions prior to 42.11.3, 43.7.0, 44.3.0, and 45.0.0-alpha.5. The flaw causes permission checks for HID and USB devices made from an iframe to incorrectly use the top-level frame's origin instead of the requesting iframe's origin. This can allow a cross-origin untrusted iframe to gain device access intended only for the top-level page. The issue is fixed in the specified versions.
### Summary
Slack Morphism exposes the correct Slack request HMAC when signature verification fails.
For an invalid request, `slack-morphism` computes the correct signature for the attacker-controlled timestamp and request body and stores it in `SlackEventWrongSignatureError.generated_hash`. The error's `Debug` and `Display` representations expose that value, and the default Axum listener error handler logs it at error level.
An attacker who can both reach the application's Slack callback endpoint and read its application/service logs can therefore use the verifier as a chosen-message HMAC oracle. The attacker can send an arbitrary request with an invalid signature, recover the correct signature from the resulting log entry, and resend the same request within the five-minute freshness window. The second request passes Slack Morphism's signature verification and reaches the protected application callback without the attacker knowing the Slack signing secret.
I reproduced this end-to-end against both the Axum and Hyper integrations in `slack-morphism` 2.26.0 and 2.27.0.
### Details
The vulnerability is in Slack Morphism's request signature verification error path.
In `src/signature_verifier.rs`, `SlackEventSignatureVerifier::verify` computes the expected signature before comparing it with the untrusted `X-Slack-Signature` supplied by the request.
On a mismatch, the implementation creates a `SlackEventWrongSignatureError` containing the complete correct signature in its public `generated_hash` field.
Relevant source path:
```text
src/signature_verifier.rs:46-59
```
The error type is defined around:
```text
src/signature_verifier.rs:182-199
```
`generated_hash` is a public `String`; the type derives `Debug`, and its `Display` implementation also prints the generated signature.
The resulting source-to-sink path is:
1. `SlackEventSignatureVerifier::verify` computes the expected HMAC for the attacker-controlled timestamp and raw request body.
2. Signature verification fails because the attacker initially sends an invalid signature.
3. Slack Morphism places the correct expected value in `SlackEventWrongSignatureError.generated_hash`.
4. The error's `Debug` and `Display` representations reveal that HMAC.
5. The Axum middleware propagates/formats the error through `src/axum_support/slack_events_middleware.rs`.
6. The default listener error handler in `src/listener.rs` logs the error at error level.
7. The attacker reads the generated HMAC from the logs.
8. The attacker resends the identical timestamp and body using the disclosed value as `X-Slack-Signature`.
9. Slack Morphism accepts the request and executes the protected callback.
The Hyper path is also affected. `src/hyper_tokio/hyper_ext.rs` propagates the verifier error as a service error, and formatting that error similarly exposes the generated signature.
The disclosed value is exactly the credential required to authenticate the corresponding request:
```text
v0=hex(HMAC-SHA256(signing_secret, "v0:" || timestamp || ":" || raw_body))
```
There is no weakness in HMAC-SHA256 or the constant-time comparison itself. The vulnerability is that the expected authentication value is disclosed through the verifier's error path.
#### Affected versions
I reproduced the vulnerability against:
- `slack-morphism` 2.26.0
- `slack-morphism` 2.27.0
- current `master` at audit time:
`e3b3ecabf99dfcf6c3548a42fba89fbfa510d21c`
Git history indicates that the public `generated_hash` field was introduced in:
```text
9bf623bd51c6a9ef0cd22d842cc707a3311f1f22
```
and appears to have existed since v0.10.0. Since versions before 2.26.0 are unsupported and were not dynamically tested during this assessment, I am reporting the independently validated supported range as:
```text
>= 2.26.0, <= 2.27.0
```
No patched release is currently available.
### PoC
The PoC uses only localhost and a synthetic signing secret.
Tested environment:
```text
slack-morphism = 2.27.0
Rust 1.98.1
Linux x86_64
Axum and Hyper integrations
```
Synthetic signing secret:
```text
SLACK_MORPHISM_SYNTHETIC_SIGNING_SECRET_7c91
```
Start a Slack Morphism Axum callback endpoint using the normal Slack request signature verification middleware and the default listener error handling.
Send a request using:
```text
Body:
{"type":"attacker_chosen","action":"transfer"}
X-Slack-Request-Timestamp:
<current Unix timestamp>
X-Slack-Signature:
v0=0000000000000000000000000000000000000000000000000000000000000000
```
The all-zero signature is deliberately invalid.
#### First request
The first request is correctly rejected with HTTP 400.
However, Slack Morphism's default error output contains an entry equivalent to:
```text
SlackEventWrongSignatureError {
...
generated_hash: "v0=<correct-HMAC>"
}
```
where `<correct-HMAC>` is the valid Slack request signature for the exact attacker-controlled timestamp and body.
Copy that `generated_hash`.
Before the five-minute timestamp window expire
Electron is a framework for writing cross-platform desktop applications using JavaScript, HTML and CSS. Prior to versions 42.11.9, 43.7.6, 44.5.0, and 45.0.0-alpha.13, pointerLock permission requests from some cross-origin iframes caused session.setPermissionRequestHandler() to receive the top-level frame's requestingUrl and isMainFrame values instead of those of the requesting frame. An application that embeds untrusted iframe content and grants pointer lock based on the requesting origin could therefore grant the permission under the top-level page's identity. Applications that deny pointerLock for untrusted embedded content are not affected. This issue is fixed in versions 42.11.9, 43.7.6, 44.5.0, and 45.0.0-alpha.13.