Re: [RFC PATCH v2 08/10] KVM: PPC: Ultravisor: Return to UV for hcalls from SVM

[Paul Mackerras <paulus@ozlabs.org>]

On Sat, May 18, 2019 at 11:25:22AM -0300, Claudio Carvalho wrote:
> From: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
> 
> All hcalls from a secure VM go to the ultravisor from where they are
> reflected into the HV. When we (HV) complete processing such hcalls,
> we should return to the UV rather than to the guest kernel.

This paragraph in the patch description, and the comment in
book3s_hv_rmhandlers.S, are confusing and possibly misleading in
focussing on returns from hcalls, when the change is needed for any
sort of entry to the guest from the hypervisor, whether it is a return
from an hcall, a return from a hypervisor interrupt, or the first time
that a guest vCPU is run.

This paragraph needs to explain that to enter a secure guest, we have
to go through the ultravisor, therefore we do a ucall when we are
entering a secure guest.

[snip]

> +/*
> + * The hcall we just completed was from Ultravisor. Use UV_RETURN
> + * ultra call to return to the Ultravisor. Results from the hcall
> + * are already in the appropriate registers (r3:12), except for
> + * R6,7 which we used as temporary registers above. Restore them,
> + * and set R0 to the ucall number (UV_RETURN).
> + */

This needs to say something like "We are entering a secure guest, so
we have to invoke the ultravisor to do that.  If we are returning from
a hcall, the results are already ...".

> +ret_to_ultra:
> +	lwz	r6, VCPU_CR(r4)
> +	mtcr	r6
> +	LOAD_REG_IMMEDIATE(r0, UV_RETURN)
> +	ld	r7, VCPU_GPR(R7)(r4)
> +	ld	r6, VCPU_GPR(R6)(r4)
> +	ld	r4, VCPU_GPR(R4)(r4)
> +	sc	2
>  
>  /*
>   * Enter the guest on a P9 or later system where we have exactly
> -- 
> 2.20.1

Paul.