CVE-2026-98275

What it actually does

Incorrect acknowledgment of RX overrun interrupts in the cortina GMAC driver due to using the wrong status register and bit positions, causing cross-port bit corruption or no-op acks on 32-bit registers.

Why it matters to the business

Potential network packet loss or degraded performance on affected Ethernet interfaces under high load, increasing operational risk for systems relying on these controllers until patched.

How an attack unfolds

An attacker generates sustained high-rate traffic to trigger RX overruns on a vulnerable cortina GMAC port; the flawed interrupt handler then either clears the wrong port's status or writes nothing, leaving the overrun condition unhandled and allowing continued or amplified disruption.

How to fix it

Update the Linux kernel to a version containing the fix that acknowledges the RX overrun bit from the correct status register (register 4) using the per-port bit.

How to confirm the fix worked

Review the gmac_irq() implementation in the cortina driver to confirm RX overrun handling references status register 4 and the matching per-port bit; test under synthetic overrun conditions to verify correct acknowledgment.

Typical attack path

1. Identify systems exposing cortina GMAC interfaces via network reconnaissance
2. Flood the interface with traffic exceeding RX buffer capacity to induce overruns
3. Observe or amplify effects from unacknowledged interrupts leading to missed packets or stalls

How to detect it

Examine kernel logs for repeated RX overrun messages paired with anomalous interrupt behavior; review dmesg or journalctl output around network interface resets or high traffic events for signs of unhandled interrupts.

What your security team should do

• Prioritize kernel updates for any systems using cortina Ethernet controllers
• Monitor upstream stable kernel releases for the merged patch
• Apply network rate limiting or monitoring on exposed interfaces until updated