CVE-2024-50157

What it actually does

The vulnerability causes an indefinite wait loop in the RDMA/bnxt_re driver for FIFO occupancy to fall below threshold after a pacing interrupt, leading to soft CPU lockups when doorbell rates are high.

Why it matters to the business

Affected systems risk processor lockups and service disruption under high RDMA or database activity; low urgency given no known exploitation or ransomware use despite broad Linux kernel presence.

How an attack unfolds

An attacker triggers the pacing interrupt then sustains a high rate of doorbell operations to force the driver into the unbounded occupancy check loop, hanging a CPU core; listed threat actors have targeted similar Linux kernel resource issues but none specifically tied to this flaw.

How to fix it

Apply the upstream fix adding a loop count limit for FPGA in __wait_for_fifo_occupancy_below_th and ensure pacing continues via bnxt_re_pacing_timer_exp; update to kernel versions containing the resolution.

How to confirm the fix worked

Inspect kernel source for the added loop count exit condition in the occupancy wait function; monitor dmesg or syslog for absence of soft lockup warnings referencing bnxt_re after high-load testing.

Typical attack path

1. Trigger pacing interrupt in bnxt_re driver
2. Sustain high doorbell rate
3. Enter unbounded __wait_for_fifo_occupancy_below_th loop
4. Induce soft lockup on processor

How to detect it

Review kernel logs and dmesg for soft lockup or hung task messages involving bnxt_re or fifo occupancy; correlate with high RDMA activity periods; no specific threat actor IOCs tied to this CVE.

What your security team should do

• Prioritize kernel updates on systems using bnxt_re RDMA
• Enable soft lockup detection and logging
• Test driver behavior under simulated high doorbell rates