Re: Bugs in multipath scsi in 4.3-rc2

[Christoph Hellwig <hch@lst.de>]

On Mon, Oct 12, 2015 at 02:45:38PM +0200, Hannes Reinecke wrote:
> You cannot easily use the driver model here as the scsi_device is
> already (potentially) bound to the ULDs.
> If you were to go with the driver model you'd have to introduce
> another sub device between scsi_target and scsi_device.

You can have two struct devices in struct scsi_device, while it's
not pretty there are plenty of example all over the kernel with
multiple devices in a single containing structure. 

> Actually I have been thinking that, as it might make my life for the
> ALUA handler easier. However, this would be quite a largish redesign
> of the current handler infrastructure, pushing out my ALUA handler
> update even more.
> So I'd like to have the ALUA changes ironed out first and merged,
> and then work on a redesigned device handler infrastructure.

Fine with me.  As mentioned before we've never supported autoloading
the device handler modules at boot time - we only ever loaded them when
explicitly attaching them through device mapper.  Waiting another
release or maybe two to finally get there isn't the end of the world.