RE: [PATCH v2 6/9] x86/hyperv: Introduce a global variable hyperv_paravisor_present

[Dexuan Cui]

> From: Dexuan Cui
> Sent: Tuesday, August 22, 2023 9:24 PM
> To: Michael Kelley (LINUX) <mikelley@xxxxxxxxxxxxx>; ak@xxxxxxxxxxxxxxx;
>  [...]
> BTW, please refer to the link for the v3 version of this patch (WIP):

I forgot to share the link in my last email... Here it is:
https://github.com/dcui/tdx/commit/3e2a28d5cf1031679bf2e2ac37eb1fd02afc8d44

> [...]
> > > --- a/drivers/hv/connection.c
> > > +++ b/drivers/hv/connection.c
> > > @@ -484,7 +484,7 @@ void vmbus_set_event(struct vmbus_channel
> > *channel)
> > >
> > >  	++channel->sig_events;
> > >
> > > -	if (hv_isolation_type_snp())
> > > +	if (hv_isolation_type_snp() && hyperv_paravisor_present)
> >
> > This code change seems to be more properly part of Patch 9 in the
> > series when hv_isolation_type_en_snp() goes away.
> 
> The change here is part of the efforts to correctly support hypercalls for
> a TDX VM with the paravisor. Patch 9 is just a clean-up patch, which is
> not really required for a TDX VM (with or with the paravisor) to run on
> Hyper-V, so IMO it's better to keep the change here in this patch.
> 
> BTW, please refer to the link for the v3 version of this patch (WIP):

https://github.com/dcui/tdx/commit/3e2a28d5cf1031679bf2e2ac37eb1fd02afc8d44
(the same link as the above one)