|   | 
      | 
  
  
      | 
      | 
  
 
     | 
    | 
  
  
     | 
    | 
  
  
    |   | 
      | 
  
  
    | 
         
xen-users
Re: [Xen-users] Advantages of HVM vs. paravirtualization in a	Linux-only
 
> Newbie here. My hardware supports HVM (full virtualization), but I only
> need to run Linux on it.  I initially tried to create a HVM domU, but
> couldn't get pciback to work with a telephony card I need for use with
> Asterisk. However everything seems to be working fine under a
> paravirtualized guest as long as I use permissive mode on the card. (Knock
> on wood-I've had some weird instability issues prior enabling permissive
> mode).
>
>
>
> Is there any real advantage to using HVM if I don't need to virtualize
> Windows?  And if there is, how do I go about passing through PCI cards to
> the HVM guest?  I'm trying to figure out if it's worth me continuing to
> hack on this problem.
No performance advantage to HVM, in that it currently incurs a necessary 
performance penalty (how much depends on exactly what you're doing).
HVM does have the advantage of closely resembling a real physical machine, 
which can be useful for some purposes.  Also handy for running other un 
xenified OSes, specific versions of Linux kernels, etc.
Also, I think that the 3.2 release provides support for /securely/[*] doing 
PCI passthrough (but only if your system has an Intel IOMMU, I think).  That 
release may not include support for using the IOMMU for PCI passthrough to PV 
domains but I know patches have been submitted so it will definitely be 
available in the future.
[*] PCI passthrough requires that you trust the domU in question with 
dom0-equivalent privilege, since it can destabilise the machine and 
potentially access all memory by abusing the DMA facility of the PCI device.
Cheers,
Mark
-- 
Push Me Pull You - Distributed SCM tool (http://www.cl.cam.ac.uk/~maw48/pmpu/)
_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users
 
 |   
 
 | 
    | 
  
  
    |   | 
    |