CVE-2021-46950

What it actually does

Failure to properly indicate failure when ending a failed write request in md/raid1 causes bitmap bits for the failed I/O to be cleared instead of being retried or marked degraded, resulting in data corruption on RAID1 arrays that use bitmaps.

Why it matters to the business

Risk of silent data corruption and loss of storage integrity on systems using Linux software RAID1 with write-intent bitmaps, increasing operational risk for storage-dependent workloads until kernels are updated.

How an attack unfolds

An attacker with the ability to induce write errors to a RAID1 device (via local access or crafted I/O) triggers the buggy failure path in raid1_end_write_request; because bitmap bits are erroneously cleared, subsequent resync or recovery operations treat the blocks as clean, allowing corruption to persist across mirrors.

How to fix it

Upgrade the Linux kernel to a version containing commit 4d856f72c10ecb060868ed10ff1b1453943fc6c8 or later; rebuild and reboot with the patched kernel; for distributions, apply the vendor-supplied kernel update that backports this fix.

How to confirm the fix worked

Confirm the running kernel contains commit 4d856f72c10ecb060868ed10ff1b1453943fc6c8 (via git log or kernel version string); test RAID1 bitmap behavior under injected write errors to verify failed requests are now correctly retried or marked degraded.

Typical attack path

1. Obtain local access or ability to trigger I/O errors on a RAID1 volume
2. Force or induce a write failure to a device in the array
3. Exploit clearing of bitmap bits in the error path
4. Allow subsequent operations to proceed with stale bitmap state leading to uncorrected corruption

How to detect it

Inspect kernel logs for raid1 write error handling messages around the time of suspected incidents; look for unexpected absence of resync activity after device failures or bitmap-related warnings in dmesg/syslog.

What your security team should do

• Prioritize patching production systems that use md/raid1 with bitmaps within the next maintenance window
• Review inventory for any remaining unpatched kernels older than the fixed commit and schedule immediate upgrades
• Enable monitoring for RAID array health and resync events as an interim detection measure