Re: [PATCH 0/2] virtio-scsi: Implement FC_HOST feature
On Mon, 01/16 09:34, Christoph Hellwig wrote: > On Tue, Jan 17, 2017 at 12:04:28AM +0800, Fam Zheng wrote: > > This series implements the proposed fc_host feature of virtio-scsi. > > Yikes. Why? Hi Christoph, this is mostly to "passthrough" a host NPIV vport (leaving libvirt doing the heavy lifting to watch and attach all visible LUNs to guest). Fam > -- > To unsubscribe from this list: send the line "unsubscribe linux-scsi" in > the body of a message to majord...@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH 0/2] virtio-scsi: Implement FC_HOST feature
> How it this supposed to work? > You do export the WWPN/WWNN of the associated host to the guest (nb: > will get interesting for non NPIV setups ...), but virtio scsi will > still do a LUN remapping. > IE the LUNs you see on the host will be different from the LUNs > presented to the guest. This is taken care of in the host by presenting to the host all LUNs from a host's NPIV vHBA. (Libvirt probably would be the one taking care of this, because QEMU may not have enough permissions). > Plus you don't _actually_ expose the FC host, but rather the WWPN of the > host presenting the LUN. > So how do you handle LUNs from different FC hosts on the guest? I'm not sure I understand. Neither I nor Fam know this stuff very well, but we are trying to do the same as Hyper-V (and other proprietary hypervisors too). > Overall, I'm not overly happy with this approach. > You already added WWPN ids to the virtio transport, so why didn't you > update the LUN field, too, to avoid this ominous LUN remapping? Is this your old idea of adding a separate target field to commands, in order to support 64-bit LUNs? That is separate, and most FC drivers only default to 16-bit LUNs anyway. > And we really should make sure to have a single FC host in the guest > presenting all LUNs. Yes, of course. Paolo
Re: [PATCH 0/2] virtio-scsi: Implement FC_HOST feature
On 01/16/2017 05:04 PM, Fam Zheng wrote: > This series implements the proposed fc_host feature of virtio-scsi. > > The first patch updates the data structure changes according to the spec > proposal; the second patch actually implements the operations. > > Fam Zheng (2): > virtio_scsi: Add fc_host definitions > virtio_scsi: Implement fc_host > > drivers/scsi/virtio_scsi.c | 55 > +++- > include/uapi/linux/virtio_scsi.h | 6 + > 2 files changed, 60 insertions(+), 1 deletion(-) > Hmm. How it this supposed to work? You do export the WWPN/WWNN of the associated host to the guest (nb: will get interesting for non NPIV setups ...), but virtio scsi will still do a LUN remapping. IE the LUNs you see on the host will be different from the LUNs presented to the guest. This makes reliable operations very tricky. Plus you don't _actually_ expose the FC host, but rather the WWPN of the host presenting the LUN. So how do you handle LUNs from different FC hosts on the guest? >From what I've seen virtio will either presenting you with with one host per LUN (so you'll end up with different SCSI hosts on the guest each having the _same_ WWPN/WWNN), or a single host presenting all LUNs, making the WWPN setting ... interesting. Overall, I'm not overly happy with this approach. You already added WWPN ids to the virtio transport, so why didn't you update the LUN field, too, to avoid this ominous LUN remapping? And we really should make sure to have a single FC host in the guest presenting all LUNs. Cheers, Hannes -- Dr. Hannes Reinecke zSeries & Storage h...@suse.de +49 911 74053 688 SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg GF: J. Hawn, J. Guild, F. Imendörffer, HRB 16746 (AG Nürnberg)
Re: [PATCH 0/2] virtio-scsi: Implement FC_HOST feature
On Tue, Jan 17, 2017 at 12:04:28AM +0800, Fam Zheng wrote: > This series implements the proposed fc_host feature of virtio-scsi. Yikes. Why?
[PATCH 0/2] virtio-scsi: Implement FC_HOST feature
This series implements the proposed fc_host feature of virtio-scsi. The first patch updates the data structure changes according to the spec proposal; the second patch actually implements the operations. Fam Zheng (2): virtio_scsi: Add fc_host definitions virtio_scsi: Implement fc_host drivers/scsi/virtio_scsi.c | 55 +++- include/uapi/linux/virtio_scsi.h | 6 + 2 files changed, 60 insertions(+), 1 deletion(-) -- 2.9.3
