Re: [PATCH 1/5] add ARMv7 HYP API to flush VM TLBs without address param

[Marc Zyngier <marc.zyngier@arm.com>]

On Thu, Apr 17 2014 at  2:33:17 am BST, Mario Smarduch <m.smarduch@samsung.com> wrote:
> Add HYP API to invalidate all VM TLBs without passing address parameter,
> that kvm_tlb_flush_vmid_ipa() uses. Hopefully this is a valid way
> to do it. Tests show nothing is broken.
>
> The address parameter is confusing since whole VM is being invalidated.
>
> Signed-off-by: Mario Smarduch <m.smarduch@samsung.com>
> ---
>  arch/arm/include/asm/kvm_asm.h  |    1 +
>  arch/arm/include/asm/kvm_host.h |    2 ++
>  arch/arm/kvm/interrupts.S       |    5 +++++
>  arch/arm/kvm/mmu.c              |    9 +++++++++
>  4 files changed, 17 insertions(+)
>
> diff --git a/arch/arm/include/asm/kvm_asm.h b/arch/arm/include/asm/kvm_asm.h
> index 661da11..090398d 100644
> --- a/arch/arm/include/asm/kvm_asm.h
> +++ b/arch/arm/include/asm/kvm_asm.h
> @@ -76,6 +76,7 @@ extern char __kvm_hyp_code_end[];
>  
>  extern void __kvm_flush_vm_context(void);
>  extern void __kvm_tlb_flush_vmid_ipa(struct kvm *kvm, phys_addr_t ipa);
> +extern void __kvm_tlb_flush_vm(struct kvm *kvm);

I'd prefer you retain the "vmid" acronym, instead of "vm". It helps
showing that we're using an invalidation by VMID instead of another
mechanism.

>  
>  extern int __kvm_vcpu_run(struct kvm_vcpu *vcpu);
>  #endif
> diff --git a/arch/arm/include/asm/kvm_host.h b/arch/arm/include/asm/kvm_host.h
> index 098f7dd..315e3f3 100644
> --- a/arch/arm/include/asm/kvm_host.h
> +++ b/arch/arm/include/asm/kvm_host.h
> @@ -228,4 +228,6 @@ int kvm_perf_teardown(void);
>  u64 kvm_arm_timer_get_reg(struct kvm_vcpu *, u64 regid);
>  int kvm_arm_timer_set_reg(struct kvm_vcpu *, u64 regid, u64 value);
>  
> +void kvm_tlb_flush_vm(struct kvm *kvm);
> +
>  #endif /* __ARM_KVM_HOST_H__ */
> diff --git a/arch/arm/kvm/interrupts.S b/arch/arm/kvm/interrupts.S
> index 0d68d40..f81c228 100644
> --- a/arch/arm/kvm/interrupts.S
> +++ b/arch/arm/kvm/interrupts.S
> @@ -45,8 +45,13 @@ __kvm_hyp_code_start:
>   *
>   * As v7 does not support flushing per IPA, just nuke the whole TLB
>   * instead, ignoring the ipa value.
> + *
> + * void __kvm_tlb_flush_vm(struct kvm *kvm) - alias on ARMv7 to flush all VM
> + * TLBs, with no need to pass IPA. Eliminate confusing code which flushes
> + * whole VM but still requires an IPA which is unused.

Loose the comment about confusion. I don't think we're that confused ;-).

>   */
>  ENTRY(__kvm_tlb_flush_vmid_ipa)
> +ENTRY(__kvm_tlb_flush_vm)
>  	push	{r2, r3}
>  
>  	dsb	ishst

Missing ENDPROC(). This may cause issues on a Thumb2 kernel, and confuse
the unwinder.

> diff --git a/arch/arm/kvm/mmu.c b/arch/arm/kvm/mmu.c
> index e8580e2..d7a1846 100644
> --- a/arch/arm/kvm/mmu.c
> +++ b/arch/arm/kvm/mmu.c
> @@ -56,6 +56,15 @@ static void kvm_tlb_flush_vmid_ipa(struct kvm *kvm, phys_addr_t ipa)
>  		kvm_call_hyp(__kvm_tlb_flush_vmid_ipa, kvm, ipa);
>  }
>  
> +/* Flushes entire VMs TLBs, for ARMv7 reuses __kvm_tlb_flush_vmid_ipa
> + * interface without the misleading address argument
> + */

Superfluous (and wrong) comment. The address argument is not misleading,
but a concious design decision.

> +void kvm_tlb_flush_vm(struct kvm *kvm)
> +{
> +	if (kvm)
> +		kvm_call_hyp(__kvm_tlb_flush_vm, kvm);
> +}
> +
>  static int mmu_topup_memory_cache(struct kvm_mmu_memory_cache *cache,
>  				  int min, int max)
>  {

-- 
Jazz is not dead. It just smells funny.