Re: [PATCH v7 10/45] x86/sev: Add support for hypervisor feature VMGEXIT

[Brijesh Singh]

On 12/2/21 11:52 AM, Borislav Petkov wrote:
> On Wed, Nov 10, 2021 at 04:06:56PM -0600, Brijesh Singh wrote:
> > +/*
> > + * The hypervisor features are available from GHCB version 2 onward.
> > + */
> > +static bool get_hv_features(void)
> > +{
> > +	u64 val;
> > +
> > +	sev_hv_features = 0;
> > +
> > +	if (ghcb_version < 2)
> > +		return false;
> > +
> > +	sev_es_wr_ghcb_msr(GHCB_MSR_HV_FT_REQ);
> > +	VMGEXIT();
> > +
> > +	val = sev_es_rd_ghcb_msr();
> > +	if (GHCB_RESP_CODE(val) != GHCB_MSR_HV_FT_RESP)
> > +		return false;
> > +
> > +	sev_hv_features = GHCB_MSR_HV_FT_RESP_VAL(val);
> > +
> > +	return true;
> > +}
>
> I still don't like this.
>
> This is more of that run-me-in-the-exception-handler thing while this is
> purely feature detection stuff which needs to be done exactly once on
> init.
>
> IOW, that stanza
>
>          if (!sev_es_negotiate_protocol())
>                  sev_es_terminate(SEV_TERM_SET_GEN, GHCB_SEV_ES_PROT_UNSUPPORTED);
>
> should be called once in sev_enable() for the decompressor kernel and
> once in sev_es_init_vc_handling() for kernel proper.

I will look into it, I will improve the sev_enable() to call GHCB MSR 
protocol query the version and feature.


> Then you don't need to do any of that sev_hv_features = 0 thing but
> detect them exactly once and query them as much as you can.
>
> Thx.