Bug 1349117

+++ This bug was initially created as a clone of Bug #1349115 +++

+++ This bug was initially created as a clone of Bug #1320621 +++

1. What is the nature and description of the request?
   As PCI Passthrough is more a pain, as all components (Hardware, Firmware, BIOS, OS) need to play well together, doing a passthrough of a NPV adapter would help in solving this issue.
   This is already done by several other vendors (IBM AIX LPAR e.g.)

2. Why does the customer need this? (List the business requirements here)
   The reason is having a virtual Fibre Channel HBA in the VM available, so that mapping storage can be easily done without the need of the RHEV admin to take actions.
   Also presenting other FC devices than disks to the VM is sometimes needed (e.g. Backup Server, which needs to handle tapes)
      
3. How would the customer like to achieve this? (List the functional requirements here)
   - Add a NPIV to a phyiscal FC-Adapter
   - Add this NPV adapter to the VM
   - The VM should be able to use the NPV adapter as if it was plain hardware
      
4. For each functional requirement listed, specify how Red Hat and the customer can test to confirm the requirement is successfully implemented.  
   Just follow the steps laidf out before. If the RHEL system running in the VM can use and access the presented devices it does work.
   http://www.ibm.com/developerworks/aix/library/au-NPIV/ is also a good source on how it should work
      
5. Is there already an existing RFE upstream or in Red Hat Bugzilla?
   Not that I am aware of. There is a similar BZ, where this has been discussed: BZ #431454
   Also see https://bugzilla.redhat.com/show_bug.cgi?id=431454#c87 for some more explanation
      
6. Does the customer have any specific timeline dependencies and which release would they like to target (i.e. RHEL5, RHEL6)?
   asap, as it is getting more difficult with each hardware generation to get a working environment.
   Also with the RMRR restrictions introduced in RHEL 7 it is impossible to do PCI passthrough with modern HP hardware.
      
7. Is the sales team involved in this request and do they have any additional input?
   No
      
8. List any affected packages or components.
   RHEV-M and qemu/libvirt
      
9. Would the customer be able to assist in testing this functionality if implemented?  
   Yes

--- Additional comment from Yaniv Kaul on 2016-03-23 16:50:51 BRT ---

Dup of bug 431454 ?

--- Additional comment from Martin Tessun on 2016-03-24 04:32:45 BRT ---

Hi Yaniv,

no, as Bug 431454 discusses a LUN passthrough with a NPV adapter being created on the hypervisor, whereas this RFE is for NPV adapter passthrough, so not a single LUN, but as discussed in Bug 431454 a complete NPV adapter.

It was asked in https://bugzilla.redhat.com/show_bug.cgi?id=431454#c92 to create a separate RFE for this.

Cheers,
Martin

--- Additional comment from Yaniv Kaul on 2016-03-24 05:11:14 BRT ---

(In reply to Martin Tessun from comment #4)
> Hi Yaniv,
> 
> no, as Bug 431454 discusses a LUN passthrough with a NPV adapter being
> created on the hypervisor, whereas this RFE is for NPV adapter passthrough,
> so not a single LUN, but as discussed in Bug 431454 a complete NPV adapter.
> 
> It was asked in https://bugzilla.redhat.com/show_bug.cgi?id=431454#c92 to
> create a separate RFE for this.

Do we know it does NOT work as regular PCI passthrough device?
> 
> Cheers,
> Martin

--- Additional comment from Martin Tessun on 2016-03-24 05:34:54 BRT ---

Hi Yaniv,

(In reply to Yaniv Kaul from comment #6)
> (In reply to Martin Tessun from comment #4)
> > Hi Yaniv,
> > 
> > no, as Bug 431454 discusses a LUN passthrough with a NPV adapter being
> > created on the hypervisor, whereas this RFE is for NPV adapter passthrough,
> > so not a single LUN, but as discussed in Bug 431454 a complete NPV adapter.
> > 
> > It was asked in https://bugzilla.redhat.com/show_bug.cgi?id=431454#c92 to
> > create a separate RFE for this.
> 
> Do we know it does NOT work as regular PCI passthrough device?

Yes, it does not work that way, as it is a virtual device, and PCI passthrough is not able to address this device.

> > 
> > Cheers,
> > Martin

--- Additional comment from Michal Skrivanek on 2016-05-26 07:47:37 BRT ---

Martin, I tried to understand the various NPIV related bugs...there are a bit too many around:) 
But IIUC this is what bug 1270581 was about. The only thing which is not there is the actual creation of vHBA, that needs to be done separately on host. But once it is there it will be seen in the hostdev UI and can be passed through as a scsi adapter into a VM
Can you confirm?

--- Additional comment from Martin Tessun on 2016-05-27 04:42:51 BRT ---

Hi Michal,

Afaik thios will not work (yet) as qemu-kvm is missing this feature. See BZ #834514 for some more details.

But as soon as this is sorted, indeed, it will probably work this way.

Cheers,
Martin