[39/59] KVM: arm64: nv: Move last_vcpu_ran to be per s2 mmu

Message ID	20190621093843.220980-40-marc.zyngier@arm.com (mailing list archive)
State	New, archived
Headers	show Return-Path: <kvm-owner@kernel.org> From: Marc Zyngier <marc.zyngier@arm.com> To: linux-arm-kernel@lists.infradead.org, kvmarm@lists.cs.columbia.edu, kvm@vger.kernel.org Cc: Andre Przywara <andre.przywara@arm.com>, Christoffer Dall <christoffer.dall@arm.com>, Dave Martin <Dave.Martin@arm.com>, Jintack Lim <jintack@cs.columbia.edu>, Julien Thierry <julien.thierry@arm.com>, James Morse <james.morse@arm.com>, Suzuki K Poulose <suzuki.poulose@arm.com> Subject: [PATCH 39/59] KVM: arm64: nv: Move last_vcpu_ran to be per s2 mmu Date: Fri, 21 Jun 2019 10:38:23 +0100 Message-Id: <20190621093843.220980-40-marc.zyngier@arm.com> In-Reply-To: <20190621093843.220980-1-marc.zyngier@arm.com> References: <20190621093843.220980-1-marc.zyngier@arm.com> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Sender: kvm-owner@vger.kernel.org Precedence: bulk
Series	KVM: arm64: ARMv8.3 Nested Virtualization support \| expand [00/59] KVM: arm64: ARMv8.3 Nested Virtualization support [01/59] KVM: arm64: Migrate _elx sysreg accessors to msr_s/mrs_s [02/59] KVM: arm64: Move __load_guest_stage2 to kvm_mmu.h [03/59] arm64: Add ARM64_HAS_NESTED_VIRT cpufeature [04/59] KVM: arm64: nv: Introduce nested virtualization VCPU feature [05/59] KVM: arm64: nv: Reset VCPU to EL2 registers if VCPU nested virt is set [06/59] KVM: arm64: nv: Allow userspace to set PSR_MODE_EL2x [07/59] KVM: arm64: nv: Add EL2 system registers to vcpu context [08/59] KVM: arm64: nv: Reset VMPIDR_EL2 and VPIDR_EL2 to sane values [09/59] KVM: arm64: nv: Add nested virt VCPU primitives for vEL2 VCPU state [10/59] KVM: arm64: nv: Support virtual EL2 exceptions [11/59] KVM: arm64: nv: Inject HVC exceptions to the virtual EL2 [12/59] KVM: arm64: nv: Handle trapped ERET from virtual EL2 [13/59] KVM: arm64: nv: Handle virtual EL2 registers in vcpu_read/write_sys_reg() [14/59] KVM: arm64: nv: Handle SPSR_EL2 specially [15/59] KVM: arm64: nv: Refactor vcpu_{read,write}_sys_reg [16/59] KVM: arm64: nv: Save/Restore vEL2 sysregs [17/59] KVM: arm64: nv: Emulate PSTATE.M for a guest hypervisor [18/59] KVM: arm64: nv: Trap EL1 VM register accesses in virtual EL2 [19/59] KVM: arm64: nv: Trap SPSR_EL1, ELR_EL1 and VBAR_EL1 from virtual EL2 [20/59] KVM: arm64: nv: Trap CPACR_EL1 access in virtual EL2 [21/59] KVM: arm64: nv: Set a handler for the system instruction traps [22/59] KVM: arm64: nv: Handle PSCI call via smc from the guest [23/59] KVM: arm64: nv: Respect virtual HCR_EL2.TWX setting [24/59] KVM: arm64: nv: Respect virtual CPTR_EL2.TFP setting [25/59] KVM: arm64: nv: Don't expose SVE to nested guests [26/59] KVM: arm64: nv: Respect the virtual HCR_EL2.NV bit setting [27/59] KVM: arm64: nv: Respect virtual HCR_EL2.TVM and TRVM settings [28/59] KVM: arm64: nv: Respect the virtual HCR_EL2.NV1 bit setting [29/59] KVM: arm64: nv: Emulate EL12 register accesses from the virtual EL2 [30/59] KVM: arm64: nv: Configure HCR_EL2 for nested virtualization [31/59] KVM: arm64: nv: Only toggle cache for virtual EL2 when SCTLR_EL2 changes [32/59] KVM: arm64: nv: Hide RAS from nested guests [33/59] KVM: arm64: nv: Pretend we only support larger-than-host page sizes [34/59] KVM: arm/arm64: nv: Factor out stage 2 page table data from struct kvm [35/59] KVM: arm/arm64: nv: Support multiple nested stage 2 mmu structures [36/59] KVM: arm64: nv: Implement nested Stage-2 page table walk logic [37/59] KVM: arm64: nv: Handle shadow stage 2 page faults [38/59] KVM: arm64: nv: Unmap/flush shadow stage 2 page tables [39/59] KVM: arm64: nv: Move last_vcpu_ran to be per s2 mmu [40/59] KVM: arm64: nv: Don't always start an S2 MMU search from the beginning [41/59] KVM: arm64: nv: Introduce sys_reg_desc.forward_trap [42/59] KVM: arm64: nv: Rework the system instruction emulation framework [43/59] KVM: arm64: nv: Trap and emulate AT instructions from virtual EL2 [44/59] KVM: arm64: nv: Trap and emulate TLBI instructions from virtual EL2 [45/59] KVM: arm64: nv: Handle traps for timer _EL02 and _EL2 sysregs accessors [46/59] KVM: arm64: nv: arch_timer: Support hyp timer emulation [47/59] KVM: arm64: nv: Propagate CNTVOFF_EL2 to the virtual EL1 timer [48/59] KVM: arm64: nv: Load timer before the GIC [49/59] KVM: arm64: nv: vgic-v3: Take cpu_if pointer directly instead of vcpu [50/59] KVM: arm64: nv: Nested GICv3 Support [51/59] KVM: arm64: nv: vgic: Emulate the HW bit in software [52/59] KVM: arm64: nv: vgic: Allow userland to set VGIC maintenance IRQ [53/59] KVM: arm64: nv: Implement maintenance interrupt forwarding [54/59] KVM: arm64: nv: Add nested GICv3 tracepoints [55/59] arm64: KVM: nv: Add handling of EL2-specific timer registers [56/59] arm64: KVM: nv: Honor SCTLR_EL2.SPAN on entering vEL2 [57/59] arm64: KVM: nv: Handle SCTLR_EL2 RES0/RES1 bits [58/59] arm64: KVM: nv: Restrict S2 RD/WR permissions to match the guest's [59/59] arm64: KVM: nv: Allow userspace to request KVM_ARM_VCPU_NESTED_VIRT

Message ID

20190621093843.220980-40-marc.zyngier@arm.com (mailing list archive)

State

New, archived

Headers

From: Marc Zyngier <marc.zyngier@arm.com>
To: linux-arm-kernel@lists.infradead.org, kvmarm@lists.cs.columbia.edu,
        kvm@vger.kernel.org
Cc: Andre Przywara <andre.przywara@arm.com>,
        Christoffer Dall <christoffer.dall@arm.com>,
        Dave Martin <Dave.Martin@arm.com>,
        Jintack Lim <jintack@cs.columbia.edu>,
        Julien Thierry <julien.thierry@arm.com>,
        James Morse <james.morse@arm.com>,
        Suzuki K Poulose <suzuki.poulose@arm.com>
Subject: [PATCH 39/59] KVM: arm64: nv: Move last_vcpu_ran to be per s2 mmu
Date: Fri, 21 Jun 2019 10:38:23 +0100
Message-Id: <20190621093843.220980-40-marc.zyngier@arm.com>
In-Reply-To: <20190621093843.220980-1-marc.zyngier@arm.com>
References: <20190621093843.220980-1-marc.zyngier@arm.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Sender: kvm-owner@vger.kernel.org
Precedence: bulk

Series

KVM: arm64: ARMv8.3 Nested Virtualization support | expand

Commit Message

Marc Zyngier June 21, 2019, 9:38 a.m. UTC

last_vcpu_ran has to be per s2 mmu now that we can have multiple S2
per VM. Let's take this opportunity to perform some cleanup.

Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
---
 arch/arm/include/asm/kvm_host.h   |  6 +++---
 arch/arm/include/asm/kvm_mmu.h    |  2 +-
 arch/arm64/include/asm/kvm_host.h |  6 +++---
 arch/arm64/include/asm/kvm_mmu.h  |  2 +-
 arch/arm64/kvm/nested.c           | 13 ++++++-------
 virt/kvm/arm/arm.c                | 22 ++++------------------
 virt/kvm/arm/mmu.c                | 26 ++++++++++++++++++++------
 7 files changed, 38 insertions(+), 39 deletions(-)

Comments

Julien Thierry July 1, 2019, 9:10 a.m. UTC | #1

On 21/06/2019 10:38, Marc Zyngier wrote:
> last_vcpu_ran has to be per s2 mmu now that we can have multiple S2
> per VM. Let's take this opportunity to perform some cleanup.
> 
> Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
> ---
>  arch/arm/include/asm/kvm_host.h   |  6 +++---
>  arch/arm/include/asm/kvm_mmu.h    |  2 +-
>  arch/arm64/include/asm/kvm_host.h |  6 +++---
>  arch/arm64/include/asm/kvm_mmu.h  |  2 +-
>  arch/arm64/kvm/nested.c           | 13 ++++++-------
>  virt/kvm/arm/arm.c                | 22 ++++------------------
>  virt/kvm/arm/mmu.c                | 26 ++++++++++++++++++++------
>  7 files changed, 38 insertions(+), 39 deletions(-)
> 
> diff --git a/arch/arm/include/asm/kvm_host.h b/arch/arm/include/asm/kvm_host.h
> index b821eb2383ad..cc761610e41e 100644
> --- a/arch/arm/include/asm/kvm_host.h
> +++ b/arch/arm/include/asm/kvm_host.h
> @@ -63,15 +63,15 @@ struct kvm_s2_mmu {
>  	pgd_t *pgd;
>  	phys_addr_t pgd_phys;
>  
> +	/* The last vcpu id that ran on each physical CPU */
> +	int __percpu *last_vcpu_ran;
> +
>  	struct kvm *kvm;
>  };
>  
>  struct kvm_arch {
>  	struct kvm_s2_mmu mmu;
>  
> -	/* The last vcpu id that ran on each physical CPU */
> -	int __percpu *last_vcpu_ran;
> -
>  	/* Stage-2 page table */
>  	pgd_t *pgd;
>  	phys_addr_t pgd_phys;
> diff --git a/arch/arm/include/asm/kvm_mmu.h b/arch/arm/include/asm/kvm_mmu.h
> index afabf1fd1d17..7a6e9008ed45 100644
> --- a/arch/arm/include/asm/kvm_mmu.h
> +++ b/arch/arm/include/asm/kvm_mmu.h
> @@ -52,7 +52,7 @@ int create_hyp_exec_mappings(phys_addr_t phys_addr, size_t size,
>  void free_hyp_pgds(void);
>  
>  void stage2_unmap_vm(struct kvm *kvm);
> -int kvm_alloc_stage2_pgd(struct kvm_s2_mmu *mmu);
> +int kvm_init_stage2_mmu(struct kvm *kvm, struct kvm_s2_mmu *mmu);
>  void kvm_free_stage2_pgd(struct kvm_s2_mmu *mmu);
>  int kvm_phys_addr_ioremap(struct kvm *kvm, phys_addr_t guest_ipa,
>  			  phys_addr_t pa, unsigned long size, bool writable);
> diff --git a/arch/arm64/include/asm/kvm_host.h b/arch/arm64/include/asm/kvm_host.h
> index cc238de170d2..b71a7a237f95 100644
> --- a/arch/arm64/include/asm/kvm_host.h
> +++ b/arch/arm64/include/asm/kvm_host.h
> @@ -104,6 +104,9 @@ struct kvm_s2_mmu {
>  	 * >0: Somebody is actively using this.
>  	 */
>  	atomic_t refcnt;
> +
> +	/* The last vcpu id that ran on each physical CPU */
> +	int __percpu *last_vcpu_ran;
>  };
>  
>  static inline bool kvm_s2_mmu_valid(struct kvm_s2_mmu *mmu)
> @@ -124,9 +127,6 @@ struct kvm_arch {
>  	/* VTCR_EL2 value for this VM */
>  	u64    vtcr;
>  
> -	/* The last vcpu id that ran on each physical CPU */
> -	int __percpu *last_vcpu_ran;
> -
>  	/* The maximum number of vCPUs depends on the used GIC model */
>  	int max_vcpus;
>  
> diff --git a/arch/arm64/include/asm/kvm_mmu.h b/arch/arm64/include/asm/kvm_mmu.h
> index f4c5ac5eb95f..53103607065a 100644
> --- a/arch/arm64/include/asm/kvm_mmu.h
> +++ b/arch/arm64/include/asm/kvm_mmu.h
> @@ -169,7 +169,7 @@ void free_hyp_pgds(void);
>  
>  void kvm_unmap_stage2_range(struct kvm_s2_mmu *mmu, phys_addr_t start, u64 size);
>  void stage2_unmap_vm(struct kvm *kvm);
> -int kvm_alloc_stage2_pgd(struct kvm_s2_mmu *mmu);
> +int kvm_init_stage2_mmu(struct kvm *kvm, struct kvm_s2_mmu *mmu);
>  void kvm_free_stage2_pgd(struct kvm_s2_mmu *mmu);
>  int kvm_phys_addr_ioremap(struct kvm *kvm, phys_addr_t guest_ipa,
>  			  phys_addr_t pa, unsigned long size, bool writable);
> diff --git a/arch/arm64/kvm/nested.c b/arch/arm64/kvm/nested.c
> index 8880033fb6e0..09afafbdc8fe 100644
> --- a/arch/arm64/kvm/nested.c
> +++ b/arch/arm64/kvm/nested.c
> @@ -52,18 +52,17 @@ int kvm_vcpu_init_nested(struct kvm_vcpu *vcpu)
>  			 GFP_KERNEL | __GFP_ZERO);
>  
>  	if (tmp) {
> -		if (tmp != kvm->arch.nested_mmus)
> +		if (tmp != kvm->arch.nested_mmus) {
>  			kfree(kvm->arch.nested_mmus);
> +			kvm->arch.nested_mmus = NULL;
> +			kvm->arch.nested_mmus_size = 0;
> +		}
>  
> -		tmp[num_mmus - 1].kvm = kvm;
> -		atomic_set(&tmp[num_mmus - 1].refcnt, 0);
> -		ret = kvm_alloc_stage2_pgd(&tmp[num_mmus - 1]);
> +		ret = kvm_init_stage2_mmu(kvm, &tmp[num_mmus - 1]);
>  		if (ret)
>  			goto out;
>  
> -		tmp[num_mmus - 2].kvm = kvm;
> -		atomic_set(&tmp[num_mmus - 2].refcnt, 0);
> -		ret = kvm_alloc_stage2_pgd(&tmp[num_mmus - 2]);
> +		ret = kvm_init_stage2_mmu(kvm, &tmp[num_mmus - 2]);
>  		if (ret) {
>  			kvm_free_stage2_pgd(&tmp[num_mmus - 1]);
>  			goto out;
> diff --git a/virt/kvm/arm/arm.c b/virt/kvm/arm/arm.c
> index bcca27d5c481..e8b584b79847 100644
> --- a/virt/kvm/arm/arm.c
> +++ b/virt/kvm/arm/arm.c
> @@ -99,29 +99,21 @@ void kvm_arch_check_processor_compat(void *rtn)
>  	*(int *)rtn = 0;
>  }
>  
> -
>  /**
>   * kvm_arch_init_vm - initializes a VM data structure
>   * @kvm:	pointer to the KVM struct
>   */
>  int kvm_arch_init_vm(struct kvm *kvm, unsigned long type)
>  {
> -	int ret, cpu;
> +	int ret;
>  
>  	ret = kvm_arm_setup_stage2(kvm, type);
>  	if (ret)
>  		return ret;
>  
> -	kvm->arch.last_vcpu_ran = alloc_percpu(typeof(*kvm->arch.last_vcpu_ran));
> -	if (!kvm->arch.last_vcpu_ran)
> -		return -ENOMEM;
> -
> -	for_each_possible_cpu(cpu)
> -		*per_cpu_ptr(kvm->arch.last_vcpu_ran, cpu) = -1;
> -
> -	ret = kvm_alloc_stage2_pgd(&kvm->arch.mmu);
> +	ret = kvm_init_stage2_mmu(kvm, &kvm->arch.mmu);
>  	if (ret)
> -		goto out_fail_alloc;
> +		return ret;
>  
>  	/* Mark the initial VMID generation invalid */
>  	kvm->arch.mmu.vmid.vmid_gen = 0;
> @@ -142,9 +134,6 @@ int kvm_arch_init_vm(struct kvm *kvm, unsigned long type)
>  	return ret;
>  out_free_stage2_pgd:
>  	kvm_free_stage2_pgd(&kvm->arch.mmu);
> -out_fail_alloc:
> -	free_percpu(kvm->arch.last_vcpu_ran);
> -	kvm->arch.last_vcpu_ran = NULL;
>  	return ret;
>  }
>  
> @@ -174,9 +163,6 @@ void kvm_arch_destroy_vm(struct kvm *kvm)
>  
>  	kvm_vgic_destroy(kvm);
>  
> -	free_percpu(kvm->arch.last_vcpu_ran);
> -	kvm->arch.last_vcpu_ran = NULL;
> -
>  	for (i = 0; i < KVM_MAX_VCPUS; ++i) {
>  		if (kvm->vcpus[i]) {
>  			kvm_arch_vcpu_free(kvm->vcpus[i]);
> @@ -359,7 +345,7 @@ void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
>  	if (nested_virt_in_use(vcpu))
>  		kvm_vcpu_load_hw_mmu(vcpu);
>  
> -	last_ran = this_cpu_ptr(vcpu->kvm->arch.last_vcpu_ran);
> +	last_ran = this_cpu_ptr(vcpu->arch.hw_mmu->last_vcpu_ran);
>  	cpu_data = this_cpu_ptr(&kvm_host_data);
>  
>  	/*
> diff --git a/virt/kvm/arm/mmu.c b/virt/kvm/arm/mmu.c
> index 94d400e7af57..6a7cba077bce 100644
> --- a/virt/kvm/arm/mmu.c
> +++ b/virt/kvm/arm/mmu.c
> @@ -903,8 +903,9 @@ int create_hyp_exec_mappings(phys_addr_t phys_addr, size_t size,
>  }
>  
>  /**
> - * kvm_alloc_stage2_pgd - allocate level-1 table for stage-2 translation.
> - * @mmu:	The stage 2 mmu struct pointer
> + * kvm_init_stage2_mmu - Initialise a S2 MMU strucrure
> + * @kvm:	The pointer to the KVM structure
> + * @mmu:	The pointer to the s2 MMU structure
>   *
>   * Allocates only the stage-2 HW PGD level table(s) of size defined by
>   * stage2_pgd_size(mmu->kvm).
> @@ -912,10 +913,11 @@ int create_hyp_exec_mappings(phys_addr_t phys_addr, size_t size,
>   * Note we don't need locking here as this is only called when the VM is
>   * created, which can only be done once.
>   */
> -int kvm_alloc_stage2_pgd(struct kvm_s2_mmu *mmu)
> +int kvm_init_stage2_mmu(struct kvm *kvm, struct kvm_s2_mmu *mmu)
>  {
>  	phys_addr_t pgd_phys;
>  	pgd_t *pgd;
> +	int cpu;
>  
>  	if (mmu->pgd != NULL) {
>  		kvm_err("kvm_arch already initialized?\n");
> @@ -923,18 +925,28 @@ int kvm_alloc_stage2_pgd(struct kvm_s2_mmu *mmu)
>  	}
>  
>  	/* Allocate the HW PGD, making sure that each page gets its own refcount */
> -	pgd = alloc_pages_exact(stage2_pgd_size(mmu->kvm), GFP_KERNEL | __GFP_ZERO);
> +	pgd = alloc_pages_exact(stage2_pgd_size(kvm), GFP_KERNEL | __GFP_ZERO);
>  	if (!pgd)
>  		return -ENOMEM;
>  
>  	pgd_phys = virt_to_phys(pgd);
> -	if (WARN_ON(pgd_phys & ~kvm_vttbr_baddr_mask(mmu->kvm)))
> +	if (WARN_ON(pgd_phys & ~kvm_vttbr_baddr_mask(kvm)))
>  		return -EINVAL;
>  
> +	mmu->last_vcpu_ran = alloc_percpu(typeof(*mmu->last_vcpu_ran));
> +	if (!mmu->last_vcpu_ran) {
> +		free_pages_exact(pgd, stage2_pgd_size(kvm));
> +		return -ENOMEM;
> +	}
> +
> +	mmu->kvm = kvm;

If we're initializing this here, we probably want to get rid of the
assignment in kvm_arch_init_vm().

>  	mmu->pgd = pgd;
>  	mmu->pgd_phys = pgd_phys;
>  	mmu->vmid.vmid_gen = 0;
>  
> +	for_each_possible_cpu(cpu)
> +		*per_cpu_ptr(mmu->last_vcpu_ran, cpu) = -1;

Nit: I'd suggest putting that right after the allocation of last_vcpu_ran.

> +
>  	kvm_init_s2_mmu(mmu);

Hmm, now we have kvm_init_stage2_mmu() and an arch (arm or arm64)
specific kvm_init_s2_mmu()...

If we want to keep the s2 mmu structure different for arm and arm64, I'd
suggest at least renaming kvm_init_s2_mmu() so the distinction with
kvm_init_stage2_mmu() is clearer.

>  
>  	return 0;
> @@ -1021,8 +1033,10 @@ void kvm_free_stage2_pgd(struct kvm_s2_mmu *mmu)
>  	spin_unlock(&kvm->mmu_lock);
>  
>  	/* Free the HW pgd, one page at a time */
> -	if (pgd)
> +	if (pgd) {
>  		free_pages_exact(pgd, stage2_pgd_size(kvm));
> +		free_percpu(mmu->last_vcpu_ran);
> +	}
>  }
>  
>  static pud_t *stage2_get_pud(struct kvm_s2_mmu *mmu, struct kvm_mmu_memory_cache *cache,
> 

Cheers,

Alexandru Elisei July 5, 2019, 3:28 p.m. UTC | #2

On 6/21/19 10:38 AM, Marc Zyngier wrote:
> last_vcpu_ran has to be per s2 mmu now that we can have multiple S2
> per VM. Let's take this opportunity to perform some cleanup.
>
> Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
> ---
>  arch/arm/include/asm/kvm_host.h   |  6 +++---
>  arch/arm/include/asm/kvm_mmu.h    |  2 +-
>  arch/arm64/include/asm/kvm_host.h |  6 +++---
>  arch/arm64/include/asm/kvm_mmu.h  |  2 +-
>  arch/arm64/kvm/nested.c           | 13 ++++++-------
>  virt/kvm/arm/arm.c                | 22 ++++------------------
>  virt/kvm/arm/mmu.c                | 26 ++++++++++++++++++++------
>  7 files changed, 38 insertions(+), 39 deletions(-)
>
> diff --git a/arch/arm/include/asm/kvm_host.h b/arch/arm/include/asm/kvm_host.h
> index b821eb2383ad..cc761610e41e 100644
> --- a/arch/arm/include/asm/kvm_host.h
> +++ b/arch/arm/include/asm/kvm_host.h
> @@ -63,15 +63,15 @@ struct kvm_s2_mmu {
>  	pgd_t *pgd;
>  	phys_addr_t pgd_phys;
>  
> +	/* The last vcpu id that ran on each physical CPU */
> +	int __percpu *last_vcpu_ran;
> +
>  	struct kvm *kvm;
>  };
>  
>  struct kvm_arch {
>  	struct kvm_s2_mmu mmu;
>  
> -	/* The last vcpu id that ran on each physical CPU */
> -	int __percpu *last_vcpu_ran;
> -
>  	/* Stage-2 page table */
>  	pgd_t *pgd;
>  	phys_addr_t pgd_phys;
> diff --git a/arch/arm/include/asm/kvm_mmu.h b/arch/arm/include/asm/kvm_mmu.h
> index afabf1fd1d17..7a6e9008ed45 100644
> --- a/arch/arm/include/asm/kvm_mmu.h
> +++ b/arch/arm/include/asm/kvm_mmu.h
> @@ -52,7 +52,7 @@ int create_hyp_exec_mappings(phys_addr_t phys_addr, size_t size,
>  void free_hyp_pgds(void);
>  
>  void stage2_unmap_vm(struct kvm *kvm);
> -int kvm_alloc_stage2_pgd(struct kvm_s2_mmu *mmu);
> +int kvm_init_stage2_mmu(struct kvm *kvm, struct kvm_s2_mmu *mmu);
>  void kvm_free_stage2_pgd(struct kvm_s2_mmu *mmu);
>  int kvm_phys_addr_ioremap(struct kvm *kvm, phys_addr_t guest_ipa,
>  			  phys_addr_t pa, unsigned long size, bool writable);
> diff --git a/arch/arm64/include/asm/kvm_host.h b/arch/arm64/include/asm/kvm_host.h
> index cc238de170d2..b71a7a237f95 100644
> --- a/arch/arm64/include/asm/kvm_host.h
> +++ b/arch/arm64/include/asm/kvm_host.h
> @@ -104,6 +104,9 @@ struct kvm_s2_mmu {
>  	 * >0: Somebody is actively using this.
>  	 */
>  	atomic_t refcnt;
> +
> +	/* The last vcpu id that ran on each physical CPU */
> +	int __percpu *last_vcpu_ran;
>  };
>  
>  static inline bool kvm_s2_mmu_valid(struct kvm_s2_mmu *mmu)
> @@ -124,9 +127,6 @@ struct kvm_arch {
>  	/* VTCR_EL2 value for this VM */
>  	u64    vtcr;
>  
> -	/* The last vcpu id that ran on each physical CPU */
> -	int __percpu *last_vcpu_ran;
> -
>  	/* The maximum number of vCPUs depends on the used GIC model */
>  	int max_vcpus;
>  
> diff --git a/arch/arm64/include/asm/kvm_mmu.h b/arch/arm64/include/asm/kvm_mmu.h
> index f4c5ac5eb95f..53103607065a 100644
> --- a/arch/arm64/include/asm/kvm_mmu.h
> +++ b/arch/arm64/include/asm/kvm_mmu.h
> @@ -169,7 +169,7 @@ void free_hyp_pgds(void);
>  
>  void kvm_unmap_stage2_range(struct kvm_s2_mmu *mmu, phys_addr_t start, u64 size);
>  void stage2_unmap_vm(struct kvm *kvm);
> -int kvm_alloc_stage2_pgd(struct kvm_s2_mmu *mmu);
> +int kvm_init_stage2_mmu(struct kvm *kvm, struct kvm_s2_mmu *mmu);
>  void kvm_free_stage2_pgd(struct kvm_s2_mmu *mmu);
>  int kvm_phys_addr_ioremap(struct kvm *kvm, phys_addr_t guest_ipa,
>  			  phys_addr_t pa, unsigned long size, bool writable);
> diff --git a/arch/arm64/kvm/nested.c b/arch/arm64/kvm/nested.c
> index 8880033fb6e0..09afafbdc8fe 100644
> --- a/arch/arm64/kvm/nested.c
> +++ b/arch/arm64/kvm/nested.c
> @@ -52,18 +52,17 @@ int kvm_vcpu_init_nested(struct kvm_vcpu *vcpu)
>  			 GFP_KERNEL | __GFP_ZERO);
>  
>  	if (tmp) {
> -		if (tmp != kvm->arch.nested_mmus)
> +		if (tmp != kvm->arch.nested_mmus) {
>  			kfree(kvm->arch.nested_mmus);
> +			kvm->arch.nested_mmus = NULL;
> +			kvm->arch.nested_mmus_size = 0;
> +		}
>  
> -		tmp[num_mmus - 1].kvm = kvm;
> -		atomic_set(&tmp[num_mmus - 1].refcnt, 0);
> -		ret = kvm_alloc_stage2_pgd(&tmp[num_mmus - 1]);
> +		ret = kvm_init_stage2_mmu(kvm, &tmp[num_mmus - 1]);
>  		if (ret)
>  			goto out;
>  
> -		tmp[num_mmus - 2].kvm = kvm;
> -		atomic_set(&tmp[num_mmus - 2].refcnt, 0);
> -		ret = kvm_alloc_stage2_pgd(&tmp[num_mmus - 2]);
> +		ret = kvm_init_stage2_mmu(kvm, &tmp[num_mmus - 2]);
>  		if (ret) {
>  			kvm_free_stage2_pgd(&tmp[num_mmus - 1]);
>  			goto out;
> diff --git a/virt/kvm/arm/arm.c b/virt/kvm/arm/arm.c
> index bcca27d5c481..e8b584b79847 100644
> --- a/virt/kvm/arm/arm.c
> +++ b/virt/kvm/arm/arm.c
> @@ -99,29 +99,21 @@ void kvm_arch_check_processor_compat(void *rtn)
>  	*(int *)rtn = 0;
>  }
>  
> -
>  /**
>   * kvm_arch_init_vm - initializes a VM data structure
>   * @kvm:	pointer to the KVM struct
>   */
>  int kvm_arch_init_vm(struct kvm *kvm, unsigned long type)
>  {
> -	int ret, cpu;
> +	int ret;
>  
>  	ret = kvm_arm_setup_stage2(kvm, type);
>  	if (ret)
>  		return ret;
>  
> -	kvm->arch.last_vcpu_ran = alloc_percpu(typeof(*kvm->arch.last_vcpu_ran));
> -	if (!kvm->arch.last_vcpu_ran)
> -		return -ENOMEM;
> -
> -	for_each_possible_cpu(cpu)
> -		*per_cpu_ptr(kvm->arch.last_vcpu_ran, cpu) = -1;
> -
> -	ret = kvm_alloc_stage2_pgd(&kvm->arch.mmu);
> +	ret = kvm_init_stage2_mmu(kvm, &kvm->arch.mmu);
>  	if (ret)
> -		goto out_fail_alloc;
> +		return ret;
>  
>  	/* Mark the initial VMID generation invalid */
>  	kvm->arch.mmu.vmid.vmid_gen = 0;

More context:

 int kvm_arch_init_vm(struct kvm *kvm, unsigned long type)
 {
-    int ret, cpu;
+    int ret;
 
     ret = kvm_arm_setup_stage2(kvm, type);
     if (ret)
         return ret;
 
-    kvm->arch.last_vcpu_ran = alloc_percpu(typeof(*kvm->arch.last_vcpu_ran));
-    if (!kvm->arch.last_vcpu_ran)
-        return -ENOMEM;
-
-    for_each_possible_cpu(cpu)
-        *per_cpu_ptr(kvm->arch.last_vcpu_ran, cpu) = -1;
-
-    ret = kvm_alloc_stage2_pgd(&kvm->arch.mmu);
+    ret = kvm_init_stage2_mmu(kvm, &kvm->arch.mmu);
     if (ret)
-        goto out_fail_alloc;
+        return ret;
 
     /* Mark the initial VMID generation invalid */
     kvm->arch.mmu.vmid.vmid_gen = 0;
     kvm->arch.mmu.kvm = kvm;
 

kvm_init_stage2_mmu already sets vmid_gen and kvm.

> @@ -142,9 +134,6 @@ int kvm_arch_init_vm(struct kvm *kvm, unsigned long type)
>  	return ret;
>  out_free_stage2_pgd:
>  	kvm_free_stage2_pgd(&kvm->arch.mmu);
> -out_fail_alloc:
> -	free_percpu(kvm->arch.last_vcpu_ran);
> -	kvm->arch.last_vcpu_ran = NULL;
>  	return ret;
>  }
>  
> @@ -174,9 +163,6 @@ void kvm_arch_destroy_vm(struct kvm *kvm)
>  
>  	kvm_vgic_destroy(kvm);
>  
> -	free_percpu(kvm->arch.last_vcpu_ran);
> -	kvm->arch.last_vcpu_ran = NULL;
> -
>  	for (i = 0; i < KVM_MAX_VCPUS; ++i) {
>  		if (kvm->vcpus[i]) {
>  			kvm_arch_vcpu_free(kvm->vcpus[i]);
> @@ -359,7 +345,7 @@ void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
>  	if (nested_virt_in_use(vcpu))
>  		kvm_vcpu_load_hw_mmu(vcpu);
>  
> -	last_ran = this_cpu_ptr(vcpu->kvm->arch.last_vcpu_ran);
> +	last_ran = this_cpu_ptr(vcpu->arch.hw_mmu->last_vcpu_ran);
>  	cpu_data = this_cpu_ptr(&kvm_host_data);
>  
>  	/*
> diff --git a/virt/kvm/arm/mmu.c b/virt/kvm/arm/mmu.c
> index 94d400e7af57..6a7cba077bce 100644
> --- a/virt/kvm/arm/mmu.c
> +++ b/virt/kvm/arm/mmu.c
> @@ -903,8 +903,9 @@ int create_hyp_exec_mappings(phys_addr_t phys_addr, size_t size,
>  }
>  
>  /**
> - * kvm_alloc_stage2_pgd - allocate level-1 table for stage-2 translation.
> - * @mmu:	The stage 2 mmu struct pointer
> + * kvm_init_stage2_mmu - Initialise a S2 MMU strucrure
> + * @kvm:	The pointer to the KVM structure
> + * @mmu:	The pointer to the s2 MMU structure
>   *
>   * Allocates only the stage-2 HW PGD level table(s) of size defined by
>   * stage2_pgd_size(mmu->kvm).
> @@ -912,10 +913,11 @@ int create_hyp_exec_mappings(phys_addr_t phys_addr, size_t size,
>   * Note we don't need locking here as this is only called when the VM is
>   * created, which can only be done once.
>   */
> -int kvm_alloc_stage2_pgd(struct kvm_s2_mmu *mmu)
> +int kvm_init_stage2_mmu(struct kvm *kvm, struct kvm_s2_mmu *mmu)
>  {
>  	phys_addr_t pgd_phys;
>  	pgd_t *pgd;
> +	int cpu;
>  
>  	if (mmu->pgd != NULL) {
>  		kvm_err("kvm_arch already initialized?\n");
> @@ -923,18 +925,28 @@ int kvm_alloc_stage2_pgd(struct kvm_s2_mmu *mmu)
>  	}
>  
>  	/* Allocate the HW PGD, making sure that each page gets its own refcount */
> -	pgd = alloc_pages_exact(stage2_pgd_size(mmu->kvm), GFP_KERNEL | __GFP_ZERO);
> +	pgd = alloc_pages_exact(stage2_pgd_size(kvm), GFP_KERNEL | __GFP_ZERO);
>  	if (!pgd)
>  		return -ENOMEM;
>  
>  	pgd_phys = virt_to_phys(pgd);
> -	if (WARN_ON(pgd_phys & ~kvm_vttbr_baddr_mask(mmu->kvm)))
> +	if (WARN_ON(pgd_phys & ~kvm_vttbr_baddr_mask(kvm)))
>  		return -EINVAL;
>  
> +	mmu->last_vcpu_ran = alloc_percpu(typeof(*mmu->last_vcpu_ran));
> +	if (!mmu->last_vcpu_ran) {
> +		free_pages_exact(pgd, stage2_pgd_size(kvm));
> +		return -ENOMEM;
> +	}
> +
> +	mmu->kvm = kvm;
>  	mmu->pgd = pgd;
>  	mmu->pgd_phys = pgd_phys;
>  	mmu->vmid.vmid_gen = 0;
>  
> +	for_each_possible_cpu(cpu)
> +		*per_cpu_ptr(mmu->last_vcpu_ran, cpu) = -1;
> +
>  	kvm_init_s2_mmu(mmu);
>  
>  	return 0;
> @@ -1021,8 +1033,10 @@ void kvm_free_stage2_pgd(struct kvm_s2_mmu *mmu)
>  	spin_unlock(&kvm->mmu_lock);
>  
>  	/* Free the HW pgd, one page at a time */

last_vcpu_ran too.

> -	if (pgd)
> +	if (pgd) {
>  		free_pages_exact(pgd, stage2_pgd_size(kvm));
> +		free_percpu(mmu->last_vcpu_ran);
> +	}
>  }
>  
>  static pud_t *stage2_get_pud(struct kvm_s2_mmu *mmu, struct kvm_mmu_memory_cache *cache,

diff --git a/arch/arm/include/asm/kvm_host.h b/arch/arm/include/asm/kvm_host.h
index b821eb2383ad..cc761610e41e 100644
--- a/arch/arm/include/asm/kvm_host.h
+++ b/arch/arm/include/asm/kvm_host.h
@@ -63,15 +63,15 @@  struct kvm_s2_mmu {
 	pgd_t *pgd;
 	phys_addr_t pgd_phys;
 
+	/* The last vcpu id that ran on each physical CPU */
+	int __percpu *last_vcpu_ran;
+
 	struct kvm *kvm;
 };
 
 struct kvm_arch {
 	struct kvm_s2_mmu mmu;
 
-	/* The last vcpu id that ran on each physical CPU */
-	int __percpu *last_vcpu_ran;
-
 	/* Stage-2 page table */
 	pgd_t *pgd;
 	phys_addr_t pgd_phys;
diff --git a/arch/arm/include/asm/kvm_mmu.h b/arch/arm/include/asm/kvm_mmu.h
index afabf1fd1d17..7a6e9008ed45 100644
--- a/arch/arm/include/asm/kvm_mmu.h
+++ b/arch/arm/include/asm/kvm_mmu.h
@@ -52,7 +52,7 @@  int create_hyp_exec_mappings(phys_addr_t phys_addr, size_t size,
 void free_hyp_pgds(void);
 
 void stage2_unmap_vm(struct kvm *kvm);
-int kvm_alloc_stage2_pgd(struct kvm_s2_mmu *mmu);
+int kvm_init_stage2_mmu(struct kvm *kvm, struct kvm_s2_mmu *mmu);
 void kvm_free_stage2_pgd(struct kvm_s2_mmu *mmu);
 int kvm_phys_addr_ioremap(struct kvm *kvm, phys_addr_t guest_ipa,
 			  phys_addr_t pa, unsigned long size, bool writable);
diff --git a/arch/arm64/include/asm/kvm_host.h b/arch/arm64/include/asm/kvm_host.h
index cc238de170d2..b71a7a237f95 100644
--- a/arch/arm64/include/asm/kvm_host.h
+++ b/arch/arm64/include/asm/kvm_host.h
@@ -104,6 +104,9 @@  struct kvm_s2_mmu {
 	 * >0: Somebody is actively using this.
 	 */
 	atomic_t refcnt;
+
+	/* The last vcpu id that ran on each physical CPU */
+	int __percpu *last_vcpu_ran;
 };
 
 static inline bool kvm_s2_mmu_valid(struct kvm_s2_mmu *mmu)
@@ -124,9 +127,6 @@  struct kvm_arch {
 	/* VTCR_EL2 value for this VM */
 	u64    vtcr;
 
-	/* The last vcpu id that ran on each physical CPU */
-	int __percpu *last_vcpu_ran;
-
 	/* The maximum number of vCPUs depends on the used GIC model */
 	int max_vcpus;
 
diff --git a/arch/arm64/include/asm/kvm_mmu.h b/arch/arm64/include/asm/kvm_mmu.h
index f4c5ac5eb95f..53103607065a 100644
--- a/arch/arm64/include/asm/kvm_mmu.h
+++ b/arch/arm64/include/asm/kvm_mmu.h
@@ -169,7 +169,7 @@  void free_hyp_pgds(void);
 
 void kvm_unmap_stage2_range(struct kvm_s2_mmu *mmu, phys_addr_t start, u64 size);
 void stage2_unmap_vm(struct kvm *kvm);
-int kvm_alloc_stage2_pgd(struct kvm_s2_mmu *mmu);
+int kvm_init_stage2_mmu(struct kvm *kvm, struct kvm_s2_mmu *mmu);
 void kvm_free_stage2_pgd(struct kvm_s2_mmu *mmu);
 int kvm_phys_addr_ioremap(struct kvm *kvm, phys_addr_t guest_ipa,
 			  phys_addr_t pa, unsigned long size, bool writable);
diff --git a/arch/arm64/kvm/nested.c b/arch/arm64/kvm/nested.c
index 8880033fb6e0..09afafbdc8fe 100644
--- a/arch/arm64/kvm/nested.c
+++ b/arch/arm64/kvm/nested.c
@@ -52,18 +52,17 @@  int kvm_vcpu_init_nested(struct kvm_vcpu *vcpu)
 			 GFP_KERNEL | __GFP_ZERO);
 
 	if (tmp) {
-		if (tmp != kvm->arch.nested_mmus)
+		if (tmp != kvm->arch.nested_mmus) {
 			kfree(kvm->arch.nested_mmus);
+			kvm->arch.nested_mmus = NULL;
+			kvm->arch.nested_mmus_size = 0;
+		}
 
-		tmp[num_mmus - 1].kvm = kvm;
-		atomic_set(&tmp[num_mmus - 1].refcnt, 0);
-		ret = kvm_alloc_stage2_pgd(&tmp[num_mmus - 1]);
+		ret = kvm_init_stage2_mmu(kvm, &tmp[num_mmus - 1]);
 		if (ret)
 			goto out;
 
-		tmp[num_mmus - 2].kvm = kvm;
-		atomic_set(&tmp[num_mmus - 2].refcnt, 0);
-		ret = kvm_alloc_stage2_pgd(&tmp[num_mmus - 2]);
+		ret = kvm_init_stage2_mmu(kvm, &tmp[num_mmus - 2]);
 		if (ret) {
 			kvm_free_stage2_pgd(&tmp[num_mmus - 1]);
 			goto out;
diff --git a/virt/kvm/arm/arm.c b/virt/kvm/arm/arm.c
index bcca27d5c481..e8b584b79847 100644
--- a/virt/kvm/arm/arm.c
+++ b/virt/kvm/arm/arm.c
@@ -99,29 +99,21 @@  void kvm_arch_check_processor_compat(void *rtn)
 	*(int *)rtn = 0;
 }
 
-
 /**
  * kvm_arch_init_vm - initializes a VM data structure
  * @kvm:	pointer to the KVM struct
  */
 int kvm_arch_init_vm(struct kvm *kvm, unsigned long type)
 {
-	int ret, cpu;
+	int ret;
 
 	ret = kvm_arm_setup_stage2(kvm, type);
 	if (ret)
 		return ret;
 
-	kvm->arch.last_vcpu_ran = alloc_percpu(typeof(*kvm->arch.last_vcpu_ran));
-	if (!kvm->arch.last_vcpu_ran)
-		return -ENOMEM;
-
-	for_each_possible_cpu(cpu)
-		*per_cpu_ptr(kvm->arch.last_vcpu_ran, cpu) = -1;
-
-	ret = kvm_alloc_stage2_pgd(&kvm->arch.mmu);
+	ret = kvm_init_stage2_mmu(kvm, &kvm->arch.mmu);
 	if (ret)
-		goto out_fail_alloc;
+		return ret;
 
 	/* Mark the initial VMID generation invalid */
 	kvm->arch.mmu.vmid.vmid_gen = 0;
@@ -142,9 +134,6 @@  int kvm_arch_init_vm(struct kvm *kvm, unsigned long type)
 	return ret;
 out_free_stage2_pgd:
 	kvm_free_stage2_pgd(&kvm->arch.mmu);
-out_fail_alloc:
-	free_percpu(kvm->arch.last_vcpu_ran);
-	kvm->arch.last_vcpu_ran = NULL;
 	return ret;
 }
 
@@ -174,9 +163,6 @@  void kvm_arch_destroy_vm(struct kvm *kvm)
 
 	kvm_vgic_destroy(kvm);
 
-	free_percpu(kvm->arch.last_vcpu_ran);
-	kvm->arch.last_vcpu_ran = NULL;
-
 	for (i = 0; i < KVM_MAX_VCPUS; ++i) {
 		if (kvm->vcpus[i]) {
 			kvm_arch_vcpu_free(kvm->vcpus[i]);
@@ -359,7 +345,7 @@  void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 	if (nested_virt_in_use(vcpu))
 		kvm_vcpu_load_hw_mmu(vcpu);
 
-	last_ran = this_cpu_ptr(vcpu->kvm->arch.last_vcpu_ran);
+	last_ran = this_cpu_ptr(vcpu->arch.hw_mmu->last_vcpu_ran);
 	cpu_data = this_cpu_ptr(&kvm_host_data);
 
 	/*
diff --git a/virt/kvm/arm/mmu.c b/virt/kvm/arm/mmu.c
index 94d400e7af57..6a7cba077bce 100644
--- a/virt/kvm/arm/mmu.c
+++ b/virt/kvm/arm/mmu.c
@@ -903,8 +903,9 @@  int create_hyp_exec_mappings(phys_addr_t phys_addr, size_t size,
 }
 
 /**
- * kvm_alloc_stage2_pgd - allocate level-1 table for stage-2 translation.
- * @mmu:	The stage 2 mmu struct pointer
+ * kvm_init_stage2_mmu - Initialise a S2 MMU strucrure
+ * @kvm:	The pointer to the KVM structure
+ * @mmu:	The pointer to the s2 MMU structure
  *
  * Allocates only the stage-2 HW PGD level table(s) of size defined by
  * stage2_pgd_size(mmu->kvm).
@@ -912,10 +913,11 @@  int create_hyp_exec_mappings(phys_addr_t phys_addr, size_t size,
  * Note we don't need locking here as this is only called when the VM is
  * created, which can only be done once.
  */
-int kvm_alloc_stage2_pgd(struct kvm_s2_mmu *mmu)
+int kvm_init_stage2_mmu(struct kvm *kvm, struct kvm_s2_mmu *mmu)
 {
 	phys_addr_t pgd_phys;
 	pgd_t *pgd;
+	int cpu;
 
 	if (mmu->pgd != NULL) {
 		kvm_err("kvm_arch already initialized?\n");
@@ -923,18 +925,28 @@  int kvm_alloc_stage2_pgd(struct kvm_s2_mmu *mmu)
 	}
 
 	/* Allocate the HW PGD, making sure that each page gets its own refcount */
-	pgd = alloc_pages_exact(stage2_pgd_size(mmu->kvm), GFP_KERNEL | __GFP_ZERO);
+	pgd = alloc_pages_exact(stage2_pgd_size(kvm), GFP_KERNEL | __GFP_ZERO);
 	if (!pgd)
 		return -ENOMEM;
 
 	pgd_phys = virt_to_phys(pgd);
-	if (WARN_ON(pgd_phys & ~kvm_vttbr_baddr_mask(mmu->kvm)))
+	if (WARN_ON(pgd_phys & ~kvm_vttbr_baddr_mask(kvm)))
 		return -EINVAL;
 
+	mmu->last_vcpu_ran = alloc_percpu(typeof(*mmu->last_vcpu_ran));
+	if (!mmu->last_vcpu_ran) {
+		free_pages_exact(pgd, stage2_pgd_size(kvm));
+		return -ENOMEM;
+	}
+
+	mmu->kvm = kvm;
 	mmu->pgd = pgd;
 	mmu->pgd_phys = pgd_phys;
 	mmu->vmid.vmid_gen = 0;
 
+	for_each_possible_cpu(cpu)
+		*per_cpu_ptr(mmu->last_vcpu_ran, cpu) = -1;
+
 	kvm_init_s2_mmu(mmu);
 
 	return 0;
@@ -1021,8 +1033,10 @@  void kvm_free_stage2_pgd(struct kvm_s2_mmu *mmu)
 	spin_unlock(&kvm->mmu_lock);
 
 	/* Free the HW pgd, one page at a time */
-	if (pgd)
+	if (pgd) {
 		free_pages_exact(pgd, stage2_pgd_size(kvm));
+		free_percpu(mmu->last_vcpu_ran);
+	}
 }
 
 static pud_t *stage2_get_pud(struct kvm_s2_mmu *mmu, struct kvm_mmu_memory_cache *cache,

[39/59] KVM: arm64: nv: Move last_vcpu_ran to be per s2 mmu

Commit Message

Comments

Patch