[ovirt-devel] virtio-scsi

[Nir Soffer]

----- Original Message -----
> From: "Marcus White" <roastedseaweed.k at gmail.com>
> To: devel at ovirt.org
> Sent: Thursday, June 12, 2014 9:49:14 PM
> Subject: Re: [ovirt-devel] virtio-scsi
> 
> Hello,
> Just a bump:) I am hoping this is the right list to ask the question...
> 
> MW.

I don't think so.

The oVirt project is build on top of libvirt and qemu. qeum developers
should be able to answer these questions.

I would try here:
http://wiki.qemu.org/MailingLists

> 
> On Fri, Jun 6, 2014 at 7:15 PM, Marcus White <roastedseaweed.k at gmail.com>
> wrote:
> > Hello
> > I had some basic questions, was asking them on the kvm list but it
> > probably makes more sense here:)
> >
> > The questions are about virtio-scsi. I would appreciate any help.
> >
> > 1. Is virtio-scsi via QEMU a zero copy operation from guest to disk?
> > The vhost-LIO as I understand it is zero copy, wasnt sure about QEMU.
> > With the VM and the QEMU thread part of the same process, it would be
> > good to know the flow and where the copies occur, if they do.
> >
> > 2. When the block layer underneath QEMU(assuming the QEMU does talk to
> > the block device and not sg, for example), if an error code comes in,
> > is it converted to a SCSI response before being passed to the guest?
> >
> > 3. Are there any other performance tweaks that would help in both the
> > QEMU case and the LIO case?
> >
> > Thanks in advance:)
> > MW
> _______________________________________________
> Devel mailing list
> Devel at ovirt.org
> http://lists.ovirt.org/mailman/listinfo/devel
>