Before we add another new error handling mechanism we need to fix the old one first. Hannes' work on not passing the scsi_cmnd to the various reset handlers hasn't made a lot of progress in the last five years and we'll need to urgently fix that first before adding even more complexity.
