UBUNTU-CVE-2026-99202: Medium severity ubuntu/linux-hwe-edge vulnerability
In the Linux kernel, the following vulnerability has been resolved: scsi: libsas: Fix SMP IO deadlock during HA resume smpexecutetasksg() calls pmruntimegetsync() on the host before issuing an SMP command. When that command is itself issued from the HA resume path, the getsync() deadlocks: it waits for the ongoing resume (the device is RPMRESUMING), while the resume is blocked in sasdrainwork() waiting for that same SMP IO to complete. The deadlock needs an expander-attached SATA disk. During sasresumeha() -> sasdrainwork(), DISCERESUME -> sasresumesata() -> atasasportresume() requests ATAEHRESET, and the hard reset for such a disk is done via SMP PHY CONTROL (sasatahardreset() -> sasphyreset() -> sassmpphycontrol() -> smpexecutetasksg()). Direct-attached SATA resets through llddcontrolphy() and SSP devices use TMFs, so neither hits this. Replace the getsync()/putsync() pair with pmruntimegetnoresume()/pmruntimeput(). smpexecutetasksg() only needs to hold off autosuspend while the SMP is in flight, and it must not try to resume the host: a sync resume issued from the HA resume path itself is what deadlocks, and by the time sasresumeha() runs, hwinit has already reinitialized the hardware, so the device is accessible without one. The usage reference is still required. Discovery work normally runs inside an event worker's PM reference, taken at sasnotifyportevent() notify time and held until the handler has flushed the disco queue. sasrediscoverexphy() however requeues DISCEREVALIDATEDOMAIN from within the revalidation worker itself, and flushworkqueue() does not wait for work items queued during execution, so that chained revalidation runs with no outer PM reference - without the getnoresume(), its SMP could race autosuspend. For the BSG path, sassmphandler() is the only caller which may find the host autosuspended: expander SMP requests do not go through any SCSI device request queue, so nothing else in that path holds the host awake. Resume it there with pmruntimeresumeandget() and check the result.
Affected Software
Remediation
Recommended actions to resolve this vulnerability, in priority order.
- Compensating control
In Linux kernel scsi/libsas, update smp_execute_task_sg() to use pm_runtime_get_noresume() and pm_runtime_put() to hold the usage reference without synchronously resuming the host; resume the host with pm_runtime_resume_and_get() and check its result before issuing the SMP command.
Event History
Frequently Asked Questions
Which systems should be prioritized for review?
Systems using an expander-attached SATA disk should be prioritized, particularly where high-availability resume operations can occur. The deadlock path requires an SMP PHY CONTROL command during the resume-time hard reset of that disk.
Are direct-attached SATA disks or SSP devices affected by this deadlock path?
No. Direct-attached SATA resets use lldd_control_phy(), while SSP devices use task management functions; neither reaches the affected SMP command path.
What behavior indicates the issue may be occurring?
The host can deadlock during HA resume: the resume path waits for SMP I/O completion in sas_drain_work(), while that SMP I/O waits for the ongoing runtime resume to finish.
What is the implemented fix?
The fix replaces pm_runtime_get_sync()/pm_runtime_put_sync() with pm_runtime_get_noresume()/pm_runtime_put(). This keeps autosuspend from occurring while SMP I/O is in flight without attempting a synchronous host resume from the HA resume path.