Re: [PATCH 07/11] KVM: arm64: Inject AArch64 exceptions from HYP

[Marc Zyngier <maz@kernel.org>]

Hi James,

On 2020-10-27 17:41, James Morse wrote:
> Hi Marc,
> 
> On 26/10/2020 13:34, Marc Zyngier wrote:
>> Move the AArch64 exception injection code from EL1 to HYP, leaving
>> only the ESR_EL1 updates to EL1. In order to come with the differences
> 
> (cope with the differences?)

Yes, much better!

>> between VHE and nVHE, two set of system register accessors are 
>> provided.
>> 
>> SPSR, ELR, PC and PSTATE are now completely handled in the hypervisor.
> 
> 
>> diff --git a/arch/arm64/kvm/hyp/exception.c 
>> b/arch/arm64/kvm/hyp/exception.c
>> index 6533a9270850..cd6e643639e8 100644
>> --- a/arch/arm64/kvm/hyp/exception.c
>> +++ b/arch/arm64/kvm/hyp/exception.c
>> @@ -11,7 +11,167 @@
>>   */
>> 
>>  #include <hyp/adjust_pc.h>
>> +#include <linux/kvm_host.h>
>> +#include <asm/kvm_emulate.h>
>> +
>> +#if defined (__KVM_NVHE_HYPERVISOR__)
>> +/*
>> + * System registers are never loaded on the CPU until we actually
>> + * restore them.
>> + */
>> +static inline u64 __vcpu_read_sys_reg(const struct kvm_vcpu *vcpu, 
>> int reg)
>> +{
>> +	return __vcpu_sys_reg(vcpu, reg);
>> +}
>> +
>> +static inline void __vcpu_write_sys_reg(struct kvm_vcpu *vcpu, u64 
>> val, int reg)
>> +{
>> +	 __vcpu_sys_reg(vcpu, reg) = val;
>> +}
>> +
>> +static void __vcpu_write_spsr(struct kvm_vcpu *vcpu, u64 val)
>> +{
>> +	write_sysreg_el1(val, SYS_SPSR);
>> +}
>> +#elif defined (__KVM_VHE_HYPERVISOR__)
>> +/* On VHE, all the registers are already loaded on the CPU */
>> +static inline u64 __vcpu_read_sys_reg(const struct kvm_vcpu *vcpu, 
>> int reg)
>> +{
>> +	u64 val;
> 
>> +	if (__vcpu_read_sys_reg_from_cpu(reg, &val))
>> +		return val;
> 
> As has_vhe()'s behaviour changes based on these KVM preprocessor 
> symbols, would:
> |	if (has_vhe() && __vcpu_read_sys_reg_from_cpu(reg, &val))
> |		return val;
> 
> let you do both of these with only one copy of the function?

Indeed that's better. Even better, let's move the has_vhe() into
__vcpu_read_sys_reg_from_cpu(), as that's the only case this is
used for.

Further cleanup could involve a new helper that would gate the
test of vcpu->sysregs_loaded_on_cpu with has_vhe() too, as this
definitely is a VHE-only feature.

> 
> 
>> +	return __vcpu_sys_reg(vcpu, reg);
>> +}
>> +
>> +static inline void __vcpu_write_sys_reg(struct kvm_vcpu *vcpu, u64 
>> val, int reg)
>> +{
>> +	if (__vcpu_write_sys_reg_to_cpu(val, reg))
>> +		return;
>> +
>> +	 __vcpu_sys_reg(vcpu, reg) = val;
>> +}
> 
> 
>> +static void __vcpu_write_spsr(struct kvm_vcpu *vcpu, u64 val)
>> +{
>> +	write_sysreg_el1(val, SYS_SPSR);
>> +}
> 
> This one doesn't look like it needs duplicating.

Spot on again, thanks!

         M.
-- 
Jazz is not dead. It just smells funny...
_______________________________________________
kvmarm mailing list
kvmarm@lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm