> + if (!lock_held) > + mutex_lock(&dev->dev_lock);
No conditional locking, please. I guess I causesd this by asking you to remove __opal_lock_unlock, but it seems we'd either need to keep it in the end. Except for that the series looks fine to me. Jens: given that 1-3 are the important fixes how about you pick those up ASAP? They all also had my Reviewed-by for previous postings.
