Re: Pe: [PATCH v5 1/3] virtio-scsi: first version

[Rusty Russell <rusty@rustcorp.com.au>]

On Sun, 12 Feb 2012 14:16:17 -0600, James Bottomley <James.Bottomley@HansenPartnership.com> wrote:
> On Thu, 2012-02-09 at 10:25 +0100, Paolo Bonzini wrote:
> > On 02/08/2012 02:37 PM, Christian Hoff wrote:
> > > Again, I have already done much testing with virtio-scsi and can confirm
> > > that the code is working flawlessly. In my opinion, virtio-scsi is a
> > > worthwhile addition to virtio-block and should be considered for inclusion
> > > into mainline kernel code.
> > 
> > Thank you very much!
> > 
> > James, will you include virtio-scsi in 3.4?
> 
> Well, no-one's yet answered the question I had about why.  virtio-scsi
> seems to be a basic duplication of virtio-blk except that it seems to
> fix some problems virtio-blk has.  Namely queue parameter discover,
> which virtio-blk doesn't seem to do.  There may also be a reason to cut
> the stack lower down.  Error handling is most often cited for this, but
> no-one's satisfactorily explaned why it's better to do error handling in
> the guest instead of the host.
> 
> Could someone please explain to me why you can't simply fix virtio-blk?
> Or would virtio-blk maintainers give a reason why they're unwilling to
> have it fixed?

My concern is simple: virtio_blk covers the 99% of cases, with very
little complexity.  To get that last 1%, we will end up re-specing much
of SCSI.

Having found someone who understand SCSI and is eager to maintain a
driver and spec, I am deeply tempted to partition the problem as simple
== virtio_blk, complex == virtio_scsi.

In fact, it would allow us to tighten the spec on VIRTIO_BLK_T_SCSI_CMD
to its actual use, which AFAICT is CDROMEJECT (maybe CDROMCLOSETRAY).

Cheers,
Rusty.