Hi Marc, On 16/02/2020 18:53, Marc Zyngier wrote: > Let's not pretend we support anything but ARMv8.0 as far as the > debug architecture is concerned. (what happens for features that disappeared?) For v8.0 the 'OS Double Lock' was mandatory. With v8.2 it became optional, and not-implemented with v8.3. The guest can see whether its implemented in ID_AA64DFR0_EL1. (and its 32bit friends) Previously these values would have at least matched, even though KVM implements it as RAZ/WI (which is the not-implemented behaviour). Would anyone care that these are inconsistent? (I've never had a solid grasp of how these debug 'lock' registers are supposed to be used). Thanks, James > diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c > index 06b2d0dc6c73..43087b50a211 100644 > --- a/arch/arm64/kvm/sys_regs.c > +++ b/arch/arm64/kvm/sys_regs.c > @@ -1094,6 +1094,9 @@ static u64 read_id_reg(const struct kvm_vcpu *vcpu, > FEATURE(ID_AA64ISAR1_GPI)); > break; > case SYS_ID_AA64DFR0_EL1: > + /* Limit debug to ARMv8.0 */ > + val &= ~FEATURE(ID_AA64DFR0_DEBUGVER); > + val |= FIELD_PREP(FEATURE(ID_AA64DFR0_DEBUGVER), 6); > /* Limit PMU to ARMv8.1 */ > val &= ~FEATURE(ID_AA64DFR0_PMUVER); > val |= FIELD_PREP(FEATURE(ID_AA64DFR0_PMUVER), 4); >
