CVE-2021-47192

What it actually does

Race condition in SCSI core sysfs state handling leads to deadlock: scsi_rescan_device holds state_mutex while scsi_send_eh_cmnd in the error handler thread blocks on the same mutex after scsi_host_in_recovery returns true, causing I/O requeuing and permanent hang during iSCSI recovery or manual state transitions.

Why it matters to the business

Storage devices become unresponsive, halting I/O operations on affected Linux systems using SCSI or iSCSI, resulting in application downtime, potential data access loss, and requiring system reboots to recover.

How an attack unfolds

An attacker with local or network access triggers SCSI device state changes via sysfs or iSCSI recovery paths while the error handler thread is active, exploiting the mutex ordering to induce deadlock and hang the host.

How to fix it

Update the Linux kernel to a version containing the fix that moves rescan code after dropping state_mutex and adds the running-state check; apply vendor-supplied kernel packages as soon as available.

How to confirm the fix worked

Confirm kernel version contains the commit moving rescan logic and the additional running-state guard; reproduce iSCSI recovery and sysfs state transitions under load to verify absence of deadlock.

Typical attack path

1. Gain local access or control iSCSI initiator to initiate device recovery or state transition
2. Invoke sysfs write to set device state to running while SCSI error handler thread is evaluating the device
3. Trigger concurrent scsi_rescan_device call that acquires state_mutex
4. Block error handler in scsi_send_eh_cmnd on the held mutex while I/O is requeued due to recovery state
5. Result in permanent hang preventing further SCSI operations

How to detect it

Review kernel logs for repeated SCSI error handler activity followed by stalled I/O and mutex-related stalls; inspect sysfs state transitions during iSCSI recovery events for signs of prolonged hangs.

What your security team should do

• Prioritize kernel updates on all systems using SCSI or iSCSI storage
• Monitor for upstream stable kernel releases containing the mutex ordering fix
• Restrict unprivileged sysfs writes to SCSI device state attributes where possible
• Test iSCSI failover and error recovery paths in staging environments after patching