RHEL6: Deadlock situation if a device removal request due to dev_loss_tmo concures with a SCAN_DEVICE Command
Issue
- We have tape devices. When tapes are added/removed, we see that devices are no longer usable. The drive seems to respond very slow sometimes.
- If a removal request and a scan request hit the SCSI Workqueue for the same rport, the workqueue gets deadlocked.
Environment
- Red Hat Enterprise Linux (RHEL) 6.5 (possibly RHEL6.6 and RHEL7 affected)
- tape drives
- Emulex HBA (8Gbit, 4Gbit and 2Gbit models affected, chip LPe12xxx and lower)
- FiberChannel (FC) target that supports Sequence Level Error Recovery
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.