Re: [PATCH v8 09/15] ARM: KVM: Memory virtualization setup

[Gleb Natapov <gleb@redhat.com>]

On Fri, Jun 15, 2012 at 03:08:22PM -0400, Christoffer Dall wrote:
> From: Christoffer Dall <cdall@cs.columbia.edu>
> 
> This commit introduces the framework for guest memory management
> through the use of 2nd stage translation. Each VM has a pointer
> to a level-1 table (the pgd field in struct kvm_arch) which is
> used for the 2nd stage translations. Entries are added when handling
> guest faults (later patch) and the table itself can be allocated and
> freed through the following functions implemented in
> arch/arm/kvm/arm_mmu.c:
>  - kvm_alloc_stage2_pgd(struct kvm *kvm);
>  - kvm_free_stage2_pgd(struct kvm *kvm);
> 
> Further, each entry in TLBs and caches are tagged with a VMID
> identifier in addition to ASIDs. The VMIDs are assigned consecutively
> to VMs in the order that VMs are executed, and caches and tlbs are
> invalidated when the VMID space has been used to allow for more than
> 255 simultaenously running guests.
> 
> The 2nd stage pgd is allocated in kvm_arch_init_vm(). The table is
> freed in kvm_arch_destroy_vm(). Both functions are called from the main
> KVM code.
> 
> Signed-off-by: Christoffer Dall <c.dall@virtualopensystems.com>
> ---
>  arch/arm/include/asm/kvm_arm.h |    2 -
>  arch/arm/include/asm/kvm_mmu.h |    5 ++
>  arch/arm/kvm/arm.c             |   65 ++++++++++++++++++++++---
>  arch/arm/kvm/mmu.c             |  103 ++++++++++++++++++++++++++++++++++++++++
>  4 files changed, 166 insertions(+), 9 deletions(-)
> 
> diff --git a/arch/arm/include/asm/kvm_arm.h b/arch/arm/include/asm/kvm_arm.h
> index 7f30cbd..257242f 100644
> --- a/arch/arm/include/asm/kvm_arm.h
> +++ b/arch/arm/include/asm/kvm_arm.h
> @@ -62,7 +62,7 @@
>   * SWIO:	Turn set/way invalidates into set/way clean+invalidate
>   */
>  #define HCR_GUEST_MASK (HCR_TSC | HCR_TWI | HCR_VM | HCR_BSU_IS | HCR_FB | \
> -			HCR_AMO | HCR_IMO | HCR_FMO | HCR_FMO | HCR_SWIO)
> +			HCR_TAC | HCR_AMO | HCR_IMO | HCR_FMO | HCR_SWIO)
>  
>  /* Hyp System Control Register (HSCTLR) bits */
>  #define HSCTLR_TE	(1 << 30)
> diff --git a/arch/arm/include/asm/kvm_mmu.h b/arch/arm/include/asm/kvm_mmu.h
> index 1aa1af4..d95662eb 100644
> --- a/arch/arm/include/asm/kvm_mmu.h
> +++ b/arch/arm/include/asm/kvm_mmu.h
> @@ -34,4 +34,9 @@ int kvm_hyp_pgd_alloc(void);
>  pgd_t *kvm_hyp_pgd_get(void);
>  void kvm_hyp_pgd_free(void);
>  
> +int kvm_alloc_stage2_pgd(struct kvm *kvm);
> +void kvm_free_stage2_pgd(struct kvm *kvm);
> +
> +int kvm_handle_guest_abort(struct kvm_vcpu *vcpu, struct kvm_run *run);
> +
>  #endif /* __ARM_KVM_MMU_H__ */
> diff --git a/arch/arm/kvm/arm.c b/arch/arm/kvm/arm.c
> index efe130c..81babe9 100644
> --- a/arch/arm/kvm/arm.c
> +++ b/arch/arm/kvm/arm.c
> @@ -38,6 +38,13 @@
>  
>  static DEFINE_PER_CPU(unsigned long, kvm_arm_hyp_stack_page);
>  
> +/* The VMID used in the VTTBR */
> +#define VMID_BITS               8
> +#define VMID_MASK               ((1 << VMID_BITS) - 1)
> +#define VMID_FIRST_GENERATION	(1 << VMID_BITS)
> +static u64 next_vmid;		/* The next available VMID in the sequence */
> +DEFINE_SPINLOCK(kvm_vmid_lock);
> +
>  int kvm_arch_hardware_enable(void *garbage)
>  {
>  	return 0;
> @@ -70,14 +77,6 @@ void kvm_arch_sync_events(struct kvm *kvm)
>  {
>  }
>  
> -int kvm_arch_init_vm(struct kvm *kvm, unsigned long type)
> -{
> -	if (type)
> -		return -EINVAL;
> -
> -	return 0;
> -}
> -
>  int kvm_arch_vcpu_fault(struct kvm_vcpu *vcpu, struct vm_fault *vmf)
>  {
>  	return VM_FAULT_SIGBUS;
> @@ -93,10 +92,46 @@ int kvm_arch_create_memslot(struct kvm_memory_slot *slot, unsigned long npages)
>  	return 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 = 0;
> +
> +	if (type)
> +		return -EINVAL;
> +
> +	ret = kvm_alloc_stage2_pgd(kvm);
> +	if (ret)
> +		goto out_fail_alloc;
> +	mutex_init(&kvm->arch.pgd_mutex);
> +
> +	ret = create_hyp_mappings(kvm, kvm + 1);
> +	if (ret)
> +		goto out_free_stage2_pgd;
> +
> +	/* Mark the initial VMID invalid */
> +	kvm->arch.vmid = 0;
> +
> +	return ret;
> +out_free_stage2_pgd:
> +	kvm_free_stage2_pgd(kvm);
> +out_fail_alloc:
> +	return ret;
> +}
> +
> +/**
> + * kvm_arch_destroy_vm - destroy the VM data structure
> + * @kvm:	pointer to the KVM struct
> + */
>  void kvm_arch_destroy_vm(struct kvm *kvm)
>  {
>  	int i;
>  
> +	kvm_free_stage2_pgd(kvm);
> +
>  	for (i = 0; i < KVM_MAX_VCPUS; ++i) {
>  		if (kvm->vcpus[i]) {
>  			kvm_arch_vcpu_free(kvm->vcpus[i]);
> @@ -172,6 +207,10 @@ struct kvm_vcpu *kvm_arch_vcpu_create(struct kvm *kvm, unsigned int id)
>  	if (err)
>  		goto free_vcpu;
>  
> +	err = create_hyp_mappings(vcpu, vcpu + 1);
> +	if (err)
> +		goto free_vcpu;
> +
>  	return vcpu;
>  free_vcpu:
>  	kmem_cache_free(kvm_vcpu_cache, vcpu);
> @@ -181,6 +220,7 @@ out:
>  
>  void kvm_arch_vcpu_free(struct kvm_vcpu *vcpu)
>  {
> +	kmem_cache_free(kvm_vcpu_cache, vcpu);
>  }
>  
>  void kvm_arch_vcpu_destroy(struct kvm_vcpu *vcpu)
> @@ -416,6 +456,15 @@ int kvm_arch_init(void *opaque)
>  	if (err)
>  		goto out_err;
>  
> +	/*
> +	 * The upper 56 bits of VMIDs are used to identify the generation
> +	 * counter, so VMIDs initialized to 0, having generation == 0, will
> +	 * never be considered valid and therefor a new VMID must always be
> +	 * assigned. Whent he VMID generation rolls over, we start from
> +	 * VMID_FIRST_GENERATION again.
> +	 */
> +	next_vmid = VMID_FIRST_GENERATION;
> +
>  	return 0;
>  out_err:
>  	return err;
> diff --git a/arch/arm/kvm/mmu.c b/arch/arm/kvm/mmu.c
> index a320b56a..b256540 100644
> --- a/arch/arm/kvm/mmu.c
> +++ b/arch/arm/kvm/mmu.c
> @@ -159,6 +159,109 @@ out:
>  	return err;
>  }
>  
> +/**
> + * kvm_alloc_stage2_pgd - allocate level-1 table for stage-2 translation.
> + * @kvm:	The KVM struct pointer for the VM.
> + *
> + * Allocates the 1st level table only of size defined by PGD2_ORDER (can
> + * support either full 40-bit input addresses or limited to 32-bit input
> + * addresses). Clears the allocated pages.
> + */
> +int kvm_alloc_stage2_pgd(struct kvm *kvm)
> +{
> +	pgd_t *pgd;
> +
> +	if (kvm->arch.pgd != NULL) {
> +		kvm_err("kvm_arch already initialized?\n");
> +		return -EINVAL;
> +	}
> +
> +	pgd = (pgd_t *)__get_free_pages(GFP_KERNEL, PGD2_ORDER);
> +	if (!pgd)
> +		return -ENOMEM;
> +
> +	memset(pgd, 0, PTRS_PER_PGD2 * sizeof(pgd_t));
> +	kvm->arch.pgd = pgd;
> +
> +	return 0;
> +}
> +
> +static void free_guest_pages(pte_t *pte, unsigned long addr)
> +{
> +	unsigned int i;
> +	struct page *page;
> +
> +	for (i = 0; i < PTRS_PER_PTE; i++, addr += PAGE_SIZE) {
Hmm, "addr" is not used.

> +		if (!pte_present(*pte))
> +			goto next_page;
Why goto instead of:

              if(pte_present(*pte)) {
> +		page = pfn_to_page(pte_pfn(*pte));
> +		put_page(page);
              }

> +next_page:
> +		pte++;
> +	}
> +}

> +
> +static void free_stage2_ptes(pmd_t *pmd, unsigned long addr)
> +{
> +	unsigned int i;
> +	pte_t *pte;
> +	struct page *page;
> +
> +	for (i = 0; i < PTRS_PER_PMD; i++, addr += PMD_SIZE) {
> +		BUG_ON(pmd_sect(*pmd));
> +		if (!pmd_none(*pmd) && pmd_table(*pmd)) {
> +			pte = pte_offset_kernel(pmd, addr);
> +			free_guest_pages(pte, addr);
> +			page = virt_to_page((void *)pte);
> +			WARN_ON(atomic_read(&page->_count) != 1);
> +			pte_free_kernel(NULL, pte);
> +		}
> +		pmd++;
> +	}
> +}
> +
> +/**
> + * kvm_free_stage2_pgd - free all stage-2 tables
> + * @kvm:	The KVM struct pointer for the VM.
> + *
> + * Walks the level-1 page table pointed to by kvm->arch.pgd and frees all
> + * underlying level-2 and level-3 tables before freeing the actual level-1 table
> + * and setting the struct pointer to NULL.
> + */
> +void kvm_free_stage2_pgd(struct kvm *kvm)
> +{
> +	pgd_t *pgd;
> +	pud_t *pud;
> +	pmd_t *pmd;
> +	unsigned long long i, addr;
> +
> +	if (kvm->arch.pgd == NULL)
> +		return;
> +
> +	/*
> +	 * We do this slightly different than other places, since we need more
> +	 * than 32 bits and for instance pgd_addr_end converts to unsigned long.
> +	 */
> +	addr = 0;
> +	for (i = 0; i < PTRS_PER_PGD2; i++) {
> +		addr = i * (unsigned long long)PGDIR_SIZE;
> +		pgd = kvm->arch.pgd + i;
> +		pud = pud_offset(pgd, addr);
> +
> +		if (pud_none(*pud))
> +			continue;
> +
> +		BUG_ON(pud_bad(*pud));
> +
> +		pmd = pmd_offset(pud, addr);
> +		free_stage2_ptes(pmd, addr);
> +		pmd_free(NULL, pmd);
> +	}
> +
> +	free_pages((unsigned long)kvm->arch.pgd, PGD2_ORDER);
> +	kvm->arch.pgd = NULL;
> +}
> +
>  int kvm_handle_guest_abort(struct kvm_vcpu *vcpu, struct kvm_run *run)
>  {
>  	return -EINVAL;
> 
> --
> To unsubscribe from this list: send the line "unsubscribe kvm" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

--
			Gleb.