On Fri, Feb 17, 2017 at 10:03:03AM +0100, Hannes Reinecke wrote: > Yes, true; it will fail. > > It will be fixed up by using reserved commands in patch 11. Patches should be fully bisectable. If that is too much work they need to be merged with a good explanation on why it is done. > > You can't rely on blk-mq being used, and we'd really want to avoid > > layering violations like this anyway. > > > Well, I _did_ enable blk-mq later on, so from that perspective, > yes, I can. No you can't. Drivers can not enable blk-mq, it's a global policy setting. > But if that's a layering violation, how am I supposed to check this? > Would be a wrapper in the scsi midlayer be acceptable? > Or is using blk_mq_tagset_busy_iter considered internal to the block layer? I think the right answer is to figure out what you're actually doing, and if it's a) nessecary at all, b) if it's already implemented in common code and c) if it should be implemented in common code.
