On Wed, Oct 21, 2020 at 08:23:54AM -0700, Stephen Boyd wrote:
> Quoting Will Deacon (2020-10-21 00:57:23)
> > On Tue, Oct 20, 2020 at 02:45:43PM -0700, Stephen Boyd wrote:
> > > According to the SMCCC spec (7.5.2 Discovery) the
> > > ARM_SMCCC_ARCH_WORKAROUND_1 function id only returns 0, 1, and
> > > SMCCC_RET_NOT_SUPPORTED corresponding to "workaround required",
> > > "workaround not required but implemented", and "who knows, you're on
> > > your own" respectively. For kvm hypercalls (hvc), we've implemented this
> > > function id to return SMCCC_RET_NOT_SUPPORTED, 1, and
> > > SMCCC_RET_NOT_REQUIRED. The SMCCC_RET_NOT_REQUIRED return value is not a
> > > thing for this function id, and is probably copy/pasted from the
> > > SMCCC_ARCH_WORKAROUND_2 function id that does support it.
> > >
> > > Clean this up by returning 0, 1, and SMCCC_RET_NOT_SUPPORTED
> > > appropriately. Changing this exposes the problem that
> > > spectre_v2_get_cpu_fw_mitigation_state() assumes a
> > > SMCCC_RET_NOT_SUPPORTED return value means we are vulnerable, but really
> > > it means we have no idea and should assume we can't do anything about
> > > mitigation. Put another way, it better be unaffected because it can't be
> > > mitigated in the firmware (in this case kvm) as the call isn't
> > > implemented!
> > >
> > > Cc: Andre Przywara <andre.przywara@xxxxxxx>
> > > Cc: Steven Price <steven.price@xxxxxxx>
> > > Cc: Marc Zyngier <maz@xxxxxxxxxx>
> > > Cc: stable@xxxxxxxxxxxxxxx
> > > Fixes: c118bbb52743 ("arm64: KVM: Propagate full Spectre v2 workaround state to KVM guests")
> > > Fixes: 73f381660959 ("arm64: Advertise mitigation of Spectre-v2, or lack thereof")
> > > Signed-off-by: Stephen Boyd <swboyd@xxxxxxxxxxxx>
> > > ---
> > >
> > > This will require a slightly different backport to stable kernels, but
> > > at least it looks like this is a problem given that this return value
> > > isn't valid per the spec and we've been going around it by returning
> > > something invalid for some time.
> > >
> > > arch/arm64/kernel/proton-pack.c | 3 +--
> > > arch/arm64/kvm/hypercalls.c | 2 +-
> > > 2 files changed, 2 insertions(+), 3 deletions(-)
> > >
> > > diff --git a/arch/arm64/kernel/proton-pack.c b/arch/arm64/kernel/proton-pack.c
> > > index 68b710f1b43f..00bd54f63f4f 100644
> > > --- a/arch/arm64/kernel/proton-pack.c
> > > +++ b/arch/arm64/kernel/proton-pack.c
> > > @@ -149,10 +149,9 @@ static enum mitigation_state spectre_v2_get_cpu_fw_mitigation_state(void)
> > > case SMCCC_RET_SUCCESS:
> > > return SPECTRE_MITIGATED;
> > > case SMCCC_ARCH_WORKAROUND_RET_UNAFFECTED:
> > > + case SMCCC_RET_NOT_SUPPORTED: /* Good luck w/ the Gatekeeper of Gozer */
> > > return SPECTRE_UNAFFECTED;
> >
> > Hmm, I'm not sure this is correct. The SMCCC spec is terrifically
> > unhelpful:
> >
> > NOT_SUPPORTED:
> > Either:
> > * None of the PEs in the system require firmware mitigation for CVE-2017-5715.
> > * The system contains at least 1 PE affected by CVE-2017-5715 that has no firmware
> > mitigation available.
> > * The firmware does not provide any information about whether firmware mitigation is
> > required.
> >
> > so we can't tell whether the thing is vulnerable or not in this case, and
> > have to assume that it is.
>
> If I'm reading the TF-A code correctly it looks like this will return
> SMC_UNK if the platform decides that "This flag can be set to 0 by the
> platform if none of the PEs in the system need the workaround." Where
> the flag is WORKAROUND_CVE_2017_5715 and the call handler returns 1 if
> the errata doesn't apply but the config is enabled, 0 if the errata
> applies and the config is enabled, or SMC_UNK (I guess this is
> NOT_SUPPORTED?) if the config is disabled[2].
>
> So TF-A could disable this config and then the kernel would think it is
> vulnerable when it actually isn't? The spec is a pile of ectoplasma
> here.
Yes, but there's not a lot we can do in that case as we rely on the
firmware to tell us whether or not we're affected. We do have the
"safelist" as a last resort, but that's about it.
> > > default:
> > > - fallthrough;
> > > - case SMCCC_RET_NOT_SUPPORTED:
> > > return SPECTRE_VULNERABLE;
> > > }
> > > }
> > > diff --git a/arch/arm64/kvm/hypercalls.c b/arch/arm64/kvm/hypercalls.c
> > > index 9824025ccc5c..868486957808 100644
> > > --- a/arch/arm64/kvm/hypercalls.c
> > > +++ b/arch/arm64/kvm/hypercalls.c
> > > @@ -31,7 +31,7 @@ int kvm_hvc_call_handler(struct kvm_vcpu *vcpu)
> > > val = SMCCC_RET_SUCCESS;
> > > break;
> > > case SPECTRE_UNAFFECTED:
> > > - val = SMCCC_RET_NOT_REQUIRED;
> > > + val = SMCCC_RET_NOT_SUPPORTED;
> >
> > Which means we need to return SMCCC_ARCH_WORKAROUND_RET_UNAFFECTED here, I
> > suppose?
> >
>
> Does the kernel implement a workaround in the case that no guest PE is
> affected? If so then returning 1 sounds OK to me, but otherwise
> NOT_SUPPORTED should work per the spec.
I don't follow you here. The spec says that "SMCCC_RET_NOT_SUPPORTED" is
valid return code in the case that "The system contains at least 1 PE
affected by CVE-2017-5715 that has no firmware mitigation available."
and do the guest would end up in the "vulnerable" state.
Will
