CVE-2026-99227
High · 7.8CVSS
7.8
EPSS
0.1%
Published
Nov 2, 2026
CISA due date
—
Cloudanix analysis
A bug in Linux could let someone who already has access to a computer cause crashes or take over the machine by confusing how certain fast input-output operations are tracked and finished.
- Exploitation
- Local Privilege
- Attacker needs
- Low Privilege
- Complexity
- Moderate
- Impact
- Full Compromise
- Component
- OS kernel
- Class
- Race Condition
- Remediation
- Upgrade
- Effort
- Medium
- Patch risk
- Low
What it actually does
The flaw allows repeated failed cancellation attempts on io_waitid requests to wrap the 31-bit reference counter to zero, enabling a later cancellation to claim ownership and complete the request a second time.
Why it matters to the business
Local attackers could crash Linux systems or escalate privileges, disrupting operations on servers and devices that rely on io_uring for high-performance I/O.
How an attack unfolds
An attacker with local access triggers multiple failed cancellations against io_uring waitid requests, causing the reference field to wrap; a subsequent cancellation then reclaims and double-completes the request, leading to memory corruption or crashes.
How to fix it
Upgrade the Linux kernel to a version containing the atomic_fetch_or change that makes cancellation idempotent for the cancel flag.
How to confirm the fix worked
Inspect kernel source for the updated __io_waitid_cancel logic using atomic_fetch_or on the cancel flag, then run targeted tests that attempt repeated cancellations without double completion.
Typical attack path
- Gain local unprivileged access on a Linux system
- Open an io_uring instance and submit waitid requests
- Repeatedly invoke cancellation paths to increment the reference counter without ownership
- Force counter wrap to zero and trigger double completion for exploitation
How to detect it
Review kernel logs, ftrace, or eBPF traces for repeated io_uring cancellation attempts on waitid requests that fail ownership checks; correlate with process activity using io_uring.
What your security team should do
- Apply kernel updates immediately on any systems exposing io_uring to untrusted local users
- Audit workloads for io_uring usage and consider disabling it via sysctl where not required
- Monitor kernel logs and traces for anomalous cancellation patterns around io_uring
The record
In the Linux kernel, the following vulnerability has been resolved: io_uring/waitid: prevent cancel reference wrap io_waitid packs the cancellation state into bit 31 of ->refs and uses bits 30:0 as a 31-bit reference counter: #define IO_WAITID_CANCEL_FLAG BIT(31) #define IO_WAITID_REF_MASK GENMASK(30, 0) __io_waitid_cancel() marks a request canceled, then increments the low bits of ->refs to claim completion ownership. When another path owns the request, the increment remains even though cancellation returns false. The request remains discoverable by cancellation, so repeated failed attempts can wrap the 31-bit reference field to zero. A later cancellation can claim ownership and complete the request a second time. Use atomic_fetch_or() when setting the cancel flag and inspect the old flag value. The first cancel attempt continues to the existing ownership increment, preserving its wake and issuer handoff semantics. Later attempts see the cancel flag and return without incrementing the reference field. This makes repeated failed cancellation attempts idempotent while leaving the first-cancel state transitions unchanged.
Affected packages
| Ecosystem | Package | Introduced | Fixed |
|---|