HIGH Introduced in 7.0
io_uring TaskRestrict Bypass
CVE-2026-99096
CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H
01Description
In the Linux kernel, the following vulnerability has been resolved: io_uring/register: don't exempt disabled rings from task restrictions A ring created with IORING_SETUP_R_DISABLED is exempt from its register restrictions until it is enabled. This lets the creator register the restrictions and then enable the ring with IORING_REGISTER_ENABLE_RINGS, whether or not that opcode is in the allowed set. Since commit ed82f35b926b ("io_uring: allow registration of per-task restrictions"), a ring also inherits restrictions from the task that creates it, and the same exemption applies to those. A task whose register opcodes are restricted can therefore create a disabled ring, issue any register opcode on it, and then enable it. Mark rings whose restrictions were inherited from the task, and keep checking register opcodes for them while the ring is disabled. Enabling such a ring now requires IORING_REGISTER_ENABLE_RINGS to be in the task's allowed set. Per-ring restrictions registered on a disabled ring work as before. [axboe: amend with inheritance flag fixup]
02KernelScan AI Analysis
Risk summary
A local unprivileged user can bypass per-task io_uring register restrictions by creating a disabled ring, which inherits the task's restrictions but incorrectly exempts them while disabled. This allows the attacker to issue arbitrary register opcodes and then enable the ring, defeating the security boundary. The vulnerability affects any system with io_uring enabled and is reachable from containers and user namespaces.
Vulnerability analysis
The io_uring subsystem lets a task limit which ring registration commands it can use. When a ring is created in a disabled state, it inherits those limits but the kernel mistakenly stopped enforcing them while the ring stayed disabled. A restricted task could therefore create a disabled ring, run any registration command it wanted — including ones its policy forbade — and then enable the ring. The fix marks rings that inherited their limits from the task and keeps enforcing those limits while the ring is disabled, so enabling the ring now requires the enable command itself to be permitted. Any local user who can call the io_uring system calls can exploit this, including processes running inside containers or user namespaces.
Lifecycle
03Fix Versions
| Branch | Introduced | Fixed in | Patch commit |
|---|---|---|---|
| 7.2 | 7.0 | 7.2.11 | f55c240192de |
| mainline | 7.0 | 7.3 | 6da8f402e362 |