Re: [PATCH] tcm_vhost: Expose ABI version via VHOST_SCSI_GET_ABI_VERSION

[Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>]

On Wed, Jul 25, 2012 at 02:14:50PM -0700, Nicholas A. Bellinger wrote:
> On Wed, 2012-07-25 at 12:55 +0100, Stefan Hajnoczi wrote:
> > On Tue, Jul 24, 2012 at 01:45:24PM -0700, Nicholas A. Bellinger wrote:
> > > On Mon, 2012-07-23 at 18:56 -0700, Greg Kroah-Hartman wrote:
> > > > On Tue, Jul 24, 2012 at 01:26:20AM +0000, Nicholas A. Bellinger wrote:
> > > > > From: Nicholas Bellinger <nab@linux-iscsi.org>
> 
> <SNIP>
> 
> > > 
> > > > > diff --git a/drivers/vhost/tcm_vhost.h b/drivers/vhost/tcm_vhost.h
> > > > > index e942df9..3d5378f 100644
> > > > > --- a/drivers/vhost/tcm_vhost.h
> > > > > +++ b/drivers/vhost/tcm_vhost.h
> > > > > @@ -80,7 +80,17 @@ struct tcm_vhost_tport {
> > > > >  
> > > > >  #include <linux/vhost.h>
> > > > >  
> > > > > +/*
> > > > > + * Used by QEMU userspace to ensure a consistent vhost-scsi ABI.
> > > > > + *
> > > > > + * ABI Rev 0: All pre 2012 revisions used by prototype out-of-tree code
> > > > > + * ABI Rev 1: 2012 version for v3.6 kernel merge candiate
> > > > > + */
> > > > > +
> > > > > +#define VHOST_SCSI_ABI_VERSION	1
> > > > > +
> > > > >  struct vhost_scsi_target {
> > > > > +	int abi_version;
> > > > >  	unsigned char vhost_wwpn[TRANSPORT_IQN_LEN];
> > > > >  	unsigned short vhost_tpgt;
> > > > >  };
> > > > > @@ -88,3 +98,4 @@ struct vhost_scsi_target {
> > > > >  /* VHOST_SCSI specific defines */
> > > > >  #define VHOST_SCSI_SET_ENDPOINT _IOW(VHOST_VIRTIO, 0x40, struct vhost_scsi_target)
> > > > >  #define VHOST_SCSI_CLEAR_ENDPOINT _IOW(VHOST_VIRTIO, 0x41, struct vhost_scsi_target)
> > > > > +#define VHOST_SCSI_GET_ABI_VERSION _IOW(VHOST_VIRTIO, 0x42, struct vhost_scsi_target)
> > > > 
> > > > No, you just broke the ABI for version "0" here, that's not how you do
> > > > this at all.
> > > > 
> > > 
> > > The intention of this patch is use ABI=1 as a starting point for
> > > tcm_vhost moving forward, with no back-wards compat for the ABI=0
> > > prototype userspace code because:
> > > 
> > > - It's based on a slightly older version of QEMU (updating the QEMU series now)
> > > - It does not have an GET_ABI_VERSION ioctl cmd (that starts with ABI=1)
> > > - It has a small user-base of target + virtio-scsi developers
> > > 
> > > So I did consider just starting from ABI=0, but figured this would help
> > > reduce the confusion for QEMU userspace wrt to the vhost-scsi code
> > > that's been floating around out-of-tree for the last 2 years.
> > 
> > There is no real user base beyond the handful of people who have hacked
> > on this.  Adding the GET_ABI_VERSION ioctl() at this stage is fine,
> > especially considering that the userspace code that talks to tcm_vhost
> > isn't in mainline in userspace yet either.
> 
> 
> Do you have a preference for a VHOST_SCSI_ABI_VERSION starting point
> here..?
> 
> I thought that v1 would be helpful to avoid confusion with the older
> userspace code, but don't really have a strong opinion either way..
> 
> Let me know what you'd prefer here, and I'll make the changes to
> tcm_vhost + vhost-scsi patch series accordingly.

I don't think 0 for out-of-tree is needed.  I'd start at 0 but either
way is okay.

The main thing I would like to confirm is that this only versions the
tcm_vhost ioctls?  In that case a single version number works.

Stefan