From mboxrd@z Thu Jan 1 00:00:00 1970 From: Bharata B Rao Date: Mon, 23 Mar 2020 04:33:14 +0000 Subject: Re: [PATCH 2/2] KVM: PPC: Book3S HV: H_SVM_INIT_START must call UV_RETURN Message-Id: <20200323042114.GH26049@in.ibm.com> List-Id: References: <20200320102643.15516-1-ldufour@linux.ibm.com> <20200320102643.15516-3-ldufour@linux.ibm.com> <20200320112403.GG26049@in.ibm.com> In-Reply-To: MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable To: Laurent Dufour Cc: linuxppc-dev@lists.ozlabs.org, linux-kernel@vger.kernel.org, kvm-ppc@vger.kernel.org, Paul Mackerras , Benjamin Herrenschmidt , Michael Ellerman On Fri, Mar 20, 2020 at 03:36:05PM +0100, Laurent Dufour wrote: > Le 20/03/2020 =E0 12:24, Bharata B Rao a =E9crit=A0: > > On Fri, Mar 20, 2020 at 11:26:43AM +0100, Laurent Dufour wrote: > > > When the call to UV_REGISTER_MEM_SLOT is failing, for instance because > > > there is not enough free secured memory, the Hypervisor (HV) has to c= all > > > UV_RETURN to report the error to the Ultravisor (UV). Then the UV wil= l call > > > H_SVM_INIT_ABORT to abort the securing phase and go back to the calli= ng VM. > > >=20 > > > If the kvm->arch.secure_guest is not set, in the return path rfid is = called > > > but there is no valid context to get back to the SVM since the Hcall = has > > > been routed by the Ultravisor. > > >=20 > > > Move the setting of kvm->arch.secure_guest earlier in > > > kvmppc_h_svm_init_start() so in the return path, UV_RETURN will be ca= lled > > > instead of rfid. > > >=20 > > > Cc: Bharata B Rao > > > Cc: Paul Mackerras > > > Cc: Benjamin Herrenschmidt > > > Cc: Michael Ellerman > > > Signed-off-by: Laurent Dufour > > > --- > > > arch/powerpc/kvm/book3s_hv_uvmem.c | 3 ++- > > > 1 file changed, 2 insertions(+), 1 deletion(-) > > >=20 > > > diff --git a/arch/powerpc/kvm/book3s_hv_uvmem.c b/arch/powerpc/kvm/bo= ok3s_hv_uvmem.c > > > index 79b1202b1c62..68dff151315c 100644 > > > --- a/arch/powerpc/kvm/book3s_hv_uvmem.c > > > +++ b/arch/powerpc/kvm/book3s_hv_uvmem.c > > > @@ -209,6 +209,8 @@ unsigned long kvmppc_h_svm_init_start(struct kvm = *kvm) > > > int ret =3D H_SUCCESS; > > > int srcu_idx; > > > + kvm->arch.secure_guest =3D KVMPPC_SECURE_INIT_START; > > > + > > > if (!kvmppc_uvmem_bitmap) > > > return H_UNSUPPORTED; > > > @@ -233,7 +235,6 @@ unsigned long kvmppc_h_svm_init_start(struct kvm = *kvm) > > > goto out; > > > } > > > } > > > - kvm->arch.secure_guest |=3D KVMPPC_SECURE_INIT_START; > >=20 > > There is an assumption that memory slots would have been registered wit= h UV > > if KVMPPC_SECURE_INIT_START has been done. KVM_PPC_SVM_OFF ioctl will s= kip > > unregistration and other steps during reboot if KVMPPC_SECURE_INIT_START > > hasn't been done. > >=20 > > Have you checked if that path isn't affected by this change? >=20 > I checked that and didn't find any issue there. >=20 > My only concern was that block: > kvm_for_each_vcpu(i, vcpu, kvm) { > spin_lock(&vcpu->arch.vpa_update_lock); > unpin_vpa_reset(kvm, &vcpu->arch.dtl); > unpin_vpa_reset(kvm, &vcpu->arch.slb_shadow); > unpin_vpa_reset(kvm, &vcpu->arch.vpa); > spin_unlock(&vcpu->arch.vpa_update_lock); > } >=20 > But that seems to be safe. Yes, looks like. >=20 > However I'm not a familiar with the KVM's code, do you think an additional > KVMPPC_SECURE_INIT_* value needed here? May be not as long as UV can handle the unexpected uv_unregister_mem_slot() calls, we are good. Regards, Bharata.