CVE-2026-98430
NoneCVSS
—
EPSS
0.2%
Published
Oct 14, 2026
CISA due date
—
Cloudanix analysis
A timing error in low-level networking hardware driver code could cause computers to crash or behave unexpectedly when starting or stopping certain high-speed network connections.
- Exploitation
- Local Privilege
- Attacker needs
- Local User
- Complexity
- Moderate
- Impact
- Denial Of Service
- Component
- OS kernel
- Class
- Race Condition
- Remediation
- Upgrade
- Effort
- Medium
- Patch risk
- Low
What it actually does
Race condition in RDMA/efa driver where IRQ handler can access admin completion and async event queues before they are fully initialized or after they have been destroyed, due to mismatched lifetimes between IRQ registration and queue construction/destruction.
Why it matters to the business
Potential system crashes or memory corruption on systems using EFA RDMA hardware during driver load/unload, leading to downtime or instability in high-performance networking environments.
How an attack unfolds
An attacker with local access times IRQ registration and queue initialization or teardown sequences to trigger the handler when resources are invalid, potentially causing denial of service or memory corruption; no known remote or in-the-wild exploitation.
How to fix it
Update to a Linux kernel version containing the fix that initializes queues with interrupts masked, registers the IRQ, then enables interrupt mode, and ensures device reset and IRQ free occur before queue destruction on removal or failure.
How to confirm the fix worked
Verify kernel source or binary contains the updated RDMA/efa initialization sequence with masked interrupts and proper teardown ordering; test driver load/unload under IRQ stress to confirm no handler access to uninitialized queues.
Typical attack path
- Gain local access to a system with EFA RDMA hardware and the vulnerable driver
- Trigger repeated module load/unload or device reset operations
- Race the IRQ handler against queue construction or destruction
- Cause invalid memory access leading to crash or corruption
How to detect it
Review kernel logs for RDMA/efa related oops, null pointer dereferences, or use-after-free during module init or removal; no specific post-exploitation artifacts known.
What your security team should do
- Prioritize kernel updates on systems with EFA hardware within standard patching windows given low exploitability indicators
- Monitor for upstream stable kernel releases containing the described fix
- Restrict unprivileged module loading where feasible until patched
The record
In the Linux kernel, the following vulnerability has been resolved: RDMA/efa: Keep admin queues alive while IRQ is registered The management IRQ handler accesses both the admin completion queue and the async event queue. The driver registered the IRQ before constructing these queues and destroyed them before freeing the IRQ, so the handler's lifetime was not contained by the resources it accesses. Initialize the queues with interrupts masked, request the IRQ, and then switch to interrupt mode. On removal, reset the device and free the IRQ before destroying the queues. Also reset the device before destroying the queues if IRQ registration fails, because the device already has their DMA addresses.
Affected packages
| Ecosystem | Package | Introduced | Fixed |
|---|