https://bugzilla.kernel.org/show_bug.cgi?id=199435
--- Comment #30 from Don ([email protected]) --- I have been re-writing the eh_reset path recently. I found a race condition between the completion handler and the reset handler. I have an update that I have been aggressively testing that seems to be holding up. I am testing using 7 volumes: 2 SAS HBAs, 2 SATA HBAs, 1 Smart Path enabled R5, and 2 LVs. My tests consist of issuing resets (sg_reset -d) to all volumes repeatedly while doing: 1. mke2fs to all 7 in parallel 2. mount all volumes in parallel 3. rsync data to all volumes in parallel 4. umount all volumes in parallel 5. fsck all volumes in parallel. Before my update, I was having the above issue. I intend to run the above tests over the weekend, then submit the update internally for code review/testing before I send the update to kernel.org. Thanks, Don Brace -- You are receiving this mail because: You are the assignee for the bug.
