Re: [Xen-devel] [PATCH v3 3/4] x86/nvmx: split updating RVI from SVI in nvmx_update_apicv

[Roger Pau Monné <roger.pau@xxxxxxxxxx>]

On Fri, Mar 27, 2020 at 02:21:46AM +0000, Tian, Kevin wrote:
> > From: Roger Pau Monne <roger.pau@xxxxxxxxxx>
> > Sent: Thursday, March 26, 2020 11:27 PM
> > 
> > Updating SVI is required when an interrupt has been injected using the
> > Ack on exit VMEXIT feature, so that the in service interrupt in the
> > GUEST_INTR_STATUS matches the vector that is signaled in
> > VM_EXIT_INTR_INFO.
> > 
> > Updating RVI however is not tied to the Ack on exit feature, as it
> > signals the next vector to be injected, and hence should always be
> > updated to the next pending vector, regardless of whether Ack on exit
> > is enabled.
> > 
> > When not using the Ack on exit feature preserve the previous vector in
> > SVI, so that it's not lost when RVI is updated to contain the pending
> > vector to inject.
> > 
> > Signed-off-by: Roger Pau Monné <roger.pau@xxxxxxxxxx>
> 
> Reviewed-by: Kevin Tian <kevin.tian@xxxxxxxxx>, with one small comment:
> 
> > ---
> > Changes since v2:
> >  - Return early if the exit reason != EXTERNAL_INTERRUPT.
> >  - Reduce the number of vmwrites by accumulating the changes to a
> >    local variable which is flushed at the end of the function.
> >  - Attempt to preserve the exiting SVI if Ack on exit is not enabled.
> > ---
> >  xen/arch/x86/hvm/vmx/vvmx.c | 33 ++++++++++++++++++++++++---------
> >  1 file changed, 24 insertions(+), 9 deletions(-)
> > 
> > diff --git a/xen/arch/x86/hvm/vmx/vvmx.c b/xen/arch/x86/hvm/vmx/vvmx.c
> > index 1753005c91..39fb553590 100644
> > --- a/xen/arch/x86/hvm/vmx/vvmx.c
> > +++ b/xen/arch/x86/hvm/vmx/vvmx.c
> > @@ -1384,28 +1384,43 @@ static void nvmx_update_apicv(struct vcpu *v)
> >      struct nestedvmx *nvmx = &vcpu_2_nvmx(v);
> >      unsigned long reason = get_vvmcs(v, VM_EXIT_REASON);
> >      unsigned long intr_info = get_vvmcs(v, VM_EXIT_INTR_INFO);
> > +    unsigned long status;
> > +    int rvi;
> > 
> > -    if ( reason == EXIT_REASON_EXTERNAL_INTERRUPT &&
> > -         nvmx->intr.source == hvm_intsrc_lapic &&
> > +    if ( reason != EXIT_REASON_EXTERNAL_INTERRUPT )
> > +        return;
> 
> can we also exit if source is not lapic? as we discussed in another
> thread, the whole logic here is only for lapic not others...

Right, in any case the code below will only update GUEST_INTR_STATUS
(either SVI or RVI) for lapic interrupts, but returning early if the
source is not the lapic will do no harm AFAICT.

Will send v4 with this changed.

Thanks, Roger.