Message-ID: <861q69oi9c.wl-maz@kernel.org>
Date: Fri, 10 May 2024 15:26:23 +0100
From: Marc Zyngier <maz@...nel.org>
To: Colton Lewis <coltonlewis@...gle.com>
Cc: kvm@...r.kernel.org, Jonathan Corbet <corbet@....net>, Oliver
 Upton <oliver.upton@...ux.dev>, James Morse <james.morse@....com>,
 Suzuki K Poulose <suzuki.poulose@....com>, Zenghui Yu <yuzenghui@...wei.com>,
 Catalin Marinas <catalin.marinas@....com>, Will Deacon <will@...nel.org>,
 linux-doc@...r.kernel.org, linux-kernel@...r.kernel.org,
 linux-arm-kernel@...ts.infradead.org, kvmarm@...ts.linux.dev
Subject: Re: [PATCH v5] KVM: arm64: Add early_param to control WFx trapping

On Tue, 30 Apr 2024 19:14:44 +0100,
Colton Lewis <coltonlewis@...gle.com> wrote:
> 
> Add an early_params to control WFI and WFE trapping. This is to
> control the degree guests can wait for interrupts on their own without
> being trapped by KVM. Options for each param are trap and notrap. trap
> enables the trap. notrap disables the trap. Absent an explicitly set
> policy, default to current behavior: disabling the trap if only a
> single task is running and enabling otherwise.
> 
> Signed-off-by: Colton Lewis <coltonlewis@...gle.com>
> ---
> v5:
> 
> * Move trap configuration to vcpu_reset_hcr(). This required moving
>   kvm_emulate.h:vcpu_reset_hcr() to arm.c:kvm_vcpu_reset_hcr() to avoid needing
>   to pull scheduler headers and my enums into kvm_emulate.h. I thought the
>   function looked too bulky for that header anyway.
> * Delete vcpu_{set,clear}_vfx_traps helpers that are no longer used anywhere.
> * Remove documentation of explicit option for default behavior to avoid any
>   implicit suggestion default behavior will stay that way.
> 
> v4:
> https://lore.kernel.org/kvmarm/20240422181716.237284-1-coltonlewis@google.com/
> 
> v3:
> https://lore.kernel.org/kvmarm/20240410175437.793508-1-coltonlewis@google.com/
> 
> v2:
> https://lore.kernel.org/kvmarm/20240319164341.1674863-1-coltonlewis@google.com/
> 
> v1:
> https://lore.kernel.org/kvmarm/20240129213918.3124494-1-coltonlewis@google.com/
> 
>  .../admin-guide/kernel-parameters.txt         |  16 +++
>  arch/arm64/include/asm/kvm_emulate.h          |  53 ---------
>  arch/arm64/include/asm/kvm_host.h             |   7 ++
>  arch/arm64/kvm/arm.c                          | 110 +++++++++++++++++-
>  4 files changed, 127 insertions(+), 59 deletions(-)
> 
> diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
> index 31b3a25680d0..a4d94d9abbe4 100644
> --- a/Documentation/admin-guide/kernel-parameters.txt
> +++ b/Documentation/admin-guide/kernel-parameters.txt
> @@ -2653,6 +2653,22 @@
>  			[KVM,ARM] Allow use of GICv4 for direct injection of
>  			LPIs.
> 
> +	kvm-arm.wfe_trap_policy=
> +			[KVM,ARM] Control when to set WFE instruction trap for
> +			KVM VMs.
> +
> +			trap: set WFE instruction trap
> +
> +			notrap: clear WFE instruction trap
> +
> +	kvm-arm.wfi_trap_policy=
> +			[KVM,ARM] Control when to set WFI instruction trap for
> +			KVM VMs.
> +
> +			trap: set WFI instruction trap
> +
> +			notrap: clear WFI instruction trap
> +

Please make it clear that neither traps are guaranteed. The
architecture *allows* an implementation to trap when no events (resp.
interrupts) are pending, but nothing more. An implementation is
perfectly allowed to ignore these bits.

>  	kvm_cma_resv_ratio=n [PPC]
>  			Reserves given percentage from system memory area for
>  			contiguous memory allocation for KVM hash pagetable
> diff --git a/arch/arm64/include/asm/kvm_emulate.h b/arch/arm64/include/asm/kvm_emulate.h
> index b804fe832184..c2a9a409ebfe 100644
> --- a/arch/arm64/include/asm/kvm_emulate.h
> +++ b/arch/arm64/include/asm/kvm_emulate.h
> @@ -67,64 +67,11 @@ static __always_inline bool vcpu_el1_is_32bit(struct kvm_vcpu *vcpu)
>  }
>  #endif
> 
> -static inline void vcpu_reset_hcr(struct kvm_vcpu *vcpu)
> -{
> -	vcpu->arch.hcr_el2 = HCR_GUEST_FLAGS;
> -	if (has_vhe() || has_hvhe())
> -		vcpu->arch.hcr_el2 |= HCR_E2H;
> -	if (cpus_have_final_cap(ARM64_HAS_RAS_EXTN)) {
> -		/* route synchronous external abort exceptions to EL2 */
> -		vcpu->arch.hcr_el2 |= HCR_TEA;
> -		/* trap error record accesses */
> -		vcpu->arch.hcr_el2 |= HCR_TERR;
> -	}
> -
> -	if (cpus_have_final_cap(ARM64_HAS_STAGE2_FWB)) {
> -		vcpu->arch.hcr_el2 |= HCR_FWB;
> -	} else {
> -		/*
> -		 * For non-FWB CPUs, we trap VM ops (HCR_EL2.TVM) until M+C
> -		 * get set in SCTLR_EL1 such that we can detect when the guest
> -		 * MMU gets turned on and do the necessary cache maintenance
> -		 * then.
> -		 */
> -		vcpu->arch.hcr_el2 |= HCR_TVM;
> -	}
> -
> -	if (cpus_have_final_cap(ARM64_HAS_EVT) &&
> -	    !cpus_have_final_cap(ARM64_MISMATCHED_CACHE_TYPE))
> -		vcpu->arch.hcr_el2 |= HCR_TID4;
> -	else
> -		vcpu->arch.hcr_el2 |= HCR_TID2;
> -
> -	if (vcpu_el1_is_32bit(vcpu))
> -		vcpu->arch.hcr_el2 &= ~HCR_RW;
> -
> -	if (kvm_has_mte(vcpu->kvm))
> -		vcpu->arch.hcr_el2 |= HCR_ATA;
> -}
> -
>  static inline unsigned long *vcpu_hcr(struct kvm_vcpu *vcpu)
>  {
>  	return (unsigned long *)&vcpu->arch.hcr_el2;
>  }
> 
> -static inline void vcpu_clear_wfx_traps(struct kvm_vcpu *vcpu)
> -{
> -	vcpu->arch.hcr_el2 &= ~HCR_TWE;
> -	if (atomic_read(&vcpu->arch.vgic_cpu.vgic_v3.its_vpe.vlpi_count) ||
> -	    vcpu->kvm->arch.vgic.nassgireq)
> -		vcpu->arch.hcr_el2 &= ~HCR_TWI;
> -	else
> -		vcpu->arch.hcr_el2 |= HCR_TWI;
> -}
> -
> -static inline void vcpu_set_wfx_traps(struct kvm_vcpu *vcpu)
> -{
> -	vcpu->arch.hcr_el2 |= HCR_TWE;
> -	vcpu->arch.hcr_el2 |= HCR_TWI;
> -}
> -
>  static inline void vcpu_ptrauth_enable(struct kvm_vcpu *vcpu)
>  {
>  	vcpu->arch.hcr_el2 |= (HCR_API | HCR_APK);
> diff --git a/arch/arm64/include/asm/kvm_host.h b/arch/arm64/include/asm/kvm_host.h
> index 21c57b812569..315ee7bfc1cb 100644
> --- a/arch/arm64/include/asm/kvm_host.h
> +++ b/arch/arm64/include/asm/kvm_host.h
> @@ -67,6 +67,13 @@ enum kvm_mode {
>  	KVM_MODE_NV,
>  	KVM_MODE_NONE,
>  };
> +
> +enum kvm_wfx_trap_policy {
> +	KVM_WFX_NOTRAP_SINGLE_TASK, /* Default option */
> +	KVM_WFX_NOTRAP,
> +	KVM_WFX_TRAP,
> +};

Since this is only ever used in arm.c, it really doesn't need to be
exposed anywhere else.

> +
>  #ifdef CONFIG_KVM
>  enum kvm_mode kvm_get_mode(void);
>  #else
> diff --git a/arch/arm64/kvm/arm.c b/arch/arm64/kvm/arm.c
> index a25265aca432..5ec52333e042 100644
> --- a/arch/arm64/kvm/arm.c
> +++ b/arch/arm64/kvm/arm.c
> @@ -46,6 +46,8 @@
>  #include <kvm/arm_psci.h>
> 
>  static enum kvm_mode kvm_mode = KVM_MODE_DEFAULT;
> +static enum kvm_wfx_trap_policy kvm_wfi_trap_policy = KVM_WFX_NOTRAP_SINGLE_TASK;
> +static enum kvm_wfx_trap_policy kvm_wfe_trap_policy = KVM_WFX_NOTRAP_SINGLE_TASK;

It would be worth declaring those as __read_mostly.

>
>  DECLARE_KVM_HYP_PER_CPU(unsigned long, kvm_hyp_vector);
> 
> @@ -456,11 +458,6 @@ void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
>  	if (kvm_arm_is_pvtime_enabled(&vcpu->arch))
>  		kvm_make_request(KVM_REQ_RECORD_STEAL, vcpu);
> 
> -	if (single_task_running())
> -		vcpu_clear_wfx_traps(vcpu);
> -	else
> -		vcpu_set_wfx_traps(vcpu);
> -
>  	if (vcpu_has_ptrauth(vcpu))
>  		vcpu_ptrauth_disable(vcpu);
>  	kvm_arch_vcpu_load_debug_state_flags(vcpu);
> @@ -1391,6 +1388,72 @@ static int kvm_vcpu_set_target(struct kvm_vcpu *vcpu,
>  	return 0;
>  }
> 
> +static bool kvm_vcpu_should_clear_twi(struct kvm_vcpu *vcpu)
> +{
> +	if (likely(kvm_wfi_trap_policy == KVM_WFX_NOTRAP_SINGLE_TASK))
> +		return single_task_running() &&
> +			(atomic_read(&vcpu->arch.vgic_cpu.vgic_v3.its_vpe.vlpi_count) ||
> +			 vcpu->kvm->arch.vgic.nassgireq);

So you are evaluating a runtime condition (scheduler queue length,
number of LPIs)...

> +
> +	return kvm_wfi_trap_policy == KVM_WFX_NOTRAP;
> +}
> +
> +static bool kvm_vcpu_should_clear_twe(struct kvm_vcpu *vcpu)
> +{
> +	if (likely(kvm_wfe_trap_policy == KVM_WFX_NOTRAP_SINGLE_TASK))
> +		return single_task_running();
> +
> +	return kvm_wfe_trap_policy == KVM_WFX_NOTRAP;
> +}
> +
> +static inline void kvm_vcpu_reset_hcr(struct kvm_vcpu *vcpu)

Why the inline?

> +{
> +	vcpu->arch.hcr_el2 = HCR_GUEST_FLAGS;
> +	if (has_vhe() || has_hvhe())
> +		vcpu->arch.hcr_el2 |= HCR_E2H;
> +	if (cpus_have_final_cap(ARM64_HAS_RAS_EXTN)) {
> +		/* route synchronous external abort exceptions to EL2 */
> +		vcpu->arch.hcr_el2 |= HCR_TEA;
> +		/* trap error record accesses */
> +		vcpu->arch.hcr_el2 |= HCR_TERR;
> +	}
> +
> +	if (cpus_have_final_cap(ARM64_HAS_STAGE2_FWB)) {
> +		vcpu->arch.hcr_el2 |= HCR_FWB;
> +	} else {
> +		/*
> +		 * For non-FWB CPUs, we trap VM ops (HCR_EL2.TVM) until M+C
> +		 * get set in SCTLR_EL1 such that we can detect when the guest
> +		 * MMU gets turned on and do the necessary cache maintenance
> +		 * then.
> +		 */
> +		vcpu->arch.hcr_el2 |= HCR_TVM;
> +	}
> +
> +	if (cpus_have_final_cap(ARM64_HAS_EVT) &&
> +	    !cpus_have_final_cap(ARM64_MISMATCHED_CACHE_TYPE))
> +		vcpu->arch.hcr_el2 |= HCR_TID4;
> +	else
> +		vcpu->arch.hcr_el2 |= HCR_TID2;
> +
> +	if (vcpu_el1_is_32bit(vcpu))
> +		vcpu->arch.hcr_el2 &= ~HCR_RW;
> +
> +	if (kvm_has_mte(vcpu->kvm))
> +		vcpu->arch.hcr_el2 |= HCR_ATA;
> +
> +
> +	if (kvm_vcpu_should_clear_twe(vcpu))
> +		vcpu->arch.hcr_el2 &= ~HCR_TWE;
> +	else
> +		vcpu->arch.hcr_el2 |= HCR_TWE;
> +
> +	if (kvm_vcpu_should_clear_twi(vcpu))
> +		vcpu->arch.hcr_el2 &= ~HCR_TWI;
> +	else
> +		vcpu->arch.hcr_el2 |= HCR_TWI;

.. and from the above runtime conditions you make it a forever
decision, for a vcpu that still hasn't executed a single instruction.
What could possibly go wrong?

> +}
> +
>  static int kvm_arch_vcpu_ioctl_vcpu_init(struct kvm_vcpu *vcpu,
>  					 struct kvm_vcpu_init *init)
>  {
> @@ -1427,7 +1490,7 @@ static int kvm_arch_vcpu_ioctl_vcpu_init(struct kvm_vcpu *vcpu,
>  			icache_inval_all_pou();
>  	}
> 
> -	vcpu_reset_hcr(vcpu);
> +	kvm_vcpu_reset_hcr(vcpu);
>  	vcpu->arch.cptr_el2 = kvm_get_reset_cptr_el2(vcpu);
> 
>  	/*
> @@ -2654,6 +2717,41 @@ static int __init early_kvm_mode_cfg(char *arg)
>  }
>  early_param("kvm-arm.mode", early_kvm_mode_cfg);
> 
> +static int __init early_kvm_wfx_trap_policy_cfg(char *arg, enum kvm_wfx_trap_policy *p)
> +{
> +	if (!arg)
> +		return -EINVAL;
> +
> +	if (strcmp(arg, "trap") == 0) {
> +		*p = KVM_WFX_TRAP;
> +		return 0;
> +	}
> +
> +	if (strcmp(arg, "notrap") == 0) {
> +		*p = KVM_WFX_NOTRAP;
> +		return 0;
> +	}
> +
> +	if (strcmp(arg, "default") == 0) {
> +		*p = KVM_WFX_NOTRAP_SINGLE_TASK;
> +		return 0;
> +	}

Where is this "default" coming from? It's not documented.

	M.

-- 
Without deviation from the norm, progress is not possible.

Powered by blists - more mailing lists