Re: [Xen-devel] [PATCH V3] x86/HVM: Introduce struct hvm_pi_ops

[Suravee Suthikulpanit <Suravee.Suthikulpanit@xxxxxxx>]

Kevin,

On 1/16/17 09:13, Tian, Kevin wrote:
> > diff --git a/xen/arch/x86/hvm/vmx/vmx.c b/xen/arch/x86/hvm/vmx/vmx.c
> > index 7b2c50c..0854e17 100644
> > --- a/xen/arch/x86/hvm/vmx/vmx.c
> > +++ b/xen/arch/x86/hvm/vmx/vmx.c
> > @@ -103,6 +103,47 @@ void vmx_pi_per_cpu_init(unsigned int cpu)
> >      spin_lock_init(&per_cpu(vmx_pi_blocking, cpu).lock);
> >  }
> >
> > +/*
> > + * To handle posted interrupts correctly, we need to set the following
> > + * state:
> > + *
> > + * * The PI notification vector (NV)
> > + * * The PI notification destination processor (NDST)
> > + * * The PI "suppress notification" bit (SN)
> > + * * The vcpu pi "blocked" list
> > + *
> > + * VMX implements the runstate transitions as the following:
> > + *
> > + * A: runnable -> running
> > + *  - SN = 0
> > + *  - NDST = v->processor
> > + *  If a VM is currently running, we want the PI delivered to the guest vcpu
> > + *  on the proper pcpu.
> > + *
> > + * B: running -> runnable
> > + *  - SN = 1
> > + *
> > + * C: running -> blocked
> > + *  - SN = 0
> > + *  - NV = pi_wakeup_vector
> > + *  - Add vcpu to blocked list
> > + *  If the vm is blocked, we want the PI delivered to Xen so that it can
> > + *  wake it up.
> > + *
> > + * D: blocked -> runnable
> > + *  - SN = 0
> > + *  - NV = posted_intr_vector
> > + *  - Take vcpu off blocked list
> > + *  If the VM is currently either preempted or offline (i.e., not running
> > + *  because of some reason other than blocking waiting for an interrupt),
> > + *  there's nothing Xen can do -- we want the interrupt pending bit set in
> > + *  the guest, but we don't want to bother Xen with an interrupt (SN clear).
> I don't know how much is common above between VMX and SVM. Can you
> post the SVM counterpart here so we can see possibility of making it a
> HVM general description?
I'm planning to put SVM counterpart in the SVM AVIC patch series. For the change
here, I'm trying to keep it as accurate as possible for the VMX. Please let me 
know
if you would like to change any wordings here.

> If we have to keep above VMX specific description, I'd prefer to putting
> it along with vmx_pi_hooks_assign.
Okay, I'll put it there.

Thanks,
Suravee

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel