Re: [PATCH v2 3/5] KVM: s390: refactor some functions in priv.c

["Nico Boehr" <nrb@linux.ibm.com>]

On Tue May 20, 2025 at 8:26 PM CEST, Claudio Imbrenda wrote:
[...]
> diff --git a/arch/s390/kvm/priv.c b/arch/s390/kvm/priv.c
> index 9253c70897a8..15843e7e57e6 100644
> --- a/arch/s390/kvm/priv.c
> +++ b/arch/s390/kvm/priv.c
[...]
> +static int skeys_common_checks(struct kvm_vcpu *vcpu, struct skeys_ops_state *state)
> +{
> +	int rc;
> +
> +	if (vcpu->arch.sie_block->gpsw.mask & PSW_MASK_PSTATE) {
> +		rc = kvm_s390_inject_program_int(vcpu, PGM_PRIVILEGED_OP);
> +		return rc ? rc : -EAGAIN;

What has changed that

return kvm_s390_inject_program_int()

is not sufficient any more?

[...]
> +	get_regs_rre_ptr(vcpu, &state->reg1, &state->reg2, &state->r1, &state->r2);
> +
> +	state->effective = vcpu->run->s.regs.gprs[state->reg2] & PAGE_MASK;

*state->r2?

[...]
>  static int handle_pfmf(struct kvm_vcpu *vcpu)
>  {
[...]
> +	if (r1.fsc) {
> +		end = kvm_s390_logical_to_effective(vcpu, end);
> +		if (kvm_s390_is_amode_64(vcpu))
> +			replace_selected_bits(r2, PAGE_MASK, end);
> +		else
> +			replace_selected_bits(r2, 0xfffff000, end);

Maybe I'm missing something, but I don't get why you need replace_selected_bits
here.  kvm_s390_logical_to_effective() already does the neccesary masking, no?