Re: [PATCH 1/3] KVM: Emulation MSI-X mask bits for assigned devices

[Sheng Yang <sheng@linux.intel.com>]

On Sunday 03 October 2010 21:03:16 Avi Kivity wrote:
>   On 09/30/2010 07:41 AM, Sheng Yang wrote:
> > On Wednesday 29 September 2010 16:36:25 Avi Kivity wrote:
> > >    On 09/28/2010 11:44 AM, Sheng Yang wrote:
> > >  >  This patch enable per-vector mask for assigned devices using MSI-X.
> > >  >  
> > >  >  +
> > >  >  
> > >  >    struct kvm_assigned_msix_entry {
> > >  >    
> > >  >    	__u32 assigned_dev_id;
> > >  >    	__u32 gsi;
> > >  >    	__u16 entry; /* The index of entry in the MSI-X table */
> > >  >  
> > >  >  -	__u16 padding[3];
> > >  >  +	__u16 flags;
> > >  >  +	__u16 padding[2];
> > >  >  
> > >  >    };
> > >  
> > >  Given that this field wasn't a flag field previously, we can't expect
> > >  it to be zero.  So before we can check it, userspace needs to tell us
> > >  that is knows the field is not padding, but a flags field.
> > >  
> > >  You can use KVM_ENABLE_CAP for that.
> > 
> > OK, I would use it. And it's per vcpu ioctl though, I would like to use
> > it on setting for per vm.
> 
> Yes.  It's easier to have KVM_ASSIGN_SET_MSIX_ENTRY2 that knows about
> the flags field.

Well, maybe it's a waste when we still have room in old one...

I think KVM_ENABLE_CAP maybe good enough.
> 
> btw, this should be documented in Documentation/kvm/api.txt.

OK.

--
regards
Yang, Sheng