Re: [PATCH v10 29/59] KVM: arm64: nv: Unmap/flush shadow stage 2 page tables

[Ganapatrao Kulkarni <gankulkarni@xxxxxxxxxxxxxxxxxxxxxx>]

Hi Marc,

On 15-05-2023 11:00 pm, Marc Zyngier wrote:
> From: Christoffer Dall <christoffer.dall@xxxxxxxxxx>
>
> Unmap/flush shadow stage 2 page tables for the nested VMs as well as the
> stage 2 page table for the guest hypervisor.
>
> Note: A bunch of the code in mmu.c relating to MMU notifiers is
> currently dealt with in an extremely abrupt way, for example by clearing
> out an entire shadow stage-2 table. This will be handled in a more
> efficient way using the reverse mapping feature in a later version of
> the patch series.

We are seeing spin-lock contention due to this patch when the 
Guest-Hypervisor(L1) is booted with higher number of cores and auto-numa 
is enabled on L0.
kvm_nested_s2_unmap is called as part of notifier call-back when numa 
page migration is happening and this function which holds lock becomes 
source of contention when there are more vCPUs are processing the 
auto-numa page fault/migration.
> Signed-off-by: Christoffer Dall <christoffer.dall@xxxxxxxxxx>
> Signed-off-by: Jintack Lim <jintack.lim@xxxxxxxxxx>
> Signed-off-by: Marc Zyngier <maz@xxxxxxxxxx>
> ---
>   arch/arm64/include/asm/kvm_mmu.h    |  3 +++
>   arch/arm64/include/asm/kvm_nested.h |  3 +++
>   arch/arm64/kvm/mmu.c                | 30 ++++++++++++++++++----
>   arch/arm64/kvm/nested.c             | 39 +++++++++++++++++++++++++++++
>   4 files changed, 70 insertions(+), 5 deletions(-)
>
> diff --git a/arch/arm64/include/asm/kvm_mmu.h b/arch/arm64/include/asm/kvm_mmu.h
> index 896acdf98e71..d155b3871c4c 100644
> --- a/arch/arm64/include/asm/kvm_mmu.h
> +++ b/arch/arm64/include/asm/kvm_mmu.h
> @@ -169,6 +169,8 @@ int create_hyp_io_mappings(phys_addr_t phys_addr, size_t size,
>   			   void __iomem **haddr);
>   int create_hyp_exec_mappings(phys_addr_t phys_addr, size_t size,
>   			     void **haddr);
> +void kvm_stage2_flush_range(struct kvm_s2_mmu *mmu,
> +			    phys_addr_t addr, phys_addr_t end);
>   void __init free_hyp_pgds(void);
>   
>   void kvm_unmap_stage2_range(struct kvm_s2_mmu *mmu, phys_addr_t start, u64 size);
> @@ -177,6 +179,7 @@ int kvm_init_stage2_mmu(struct kvm *kvm, struct kvm_s2_mmu *mmu, unsigned long t
>   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);
> +void kvm_stage2_wp_range(struct kvm_s2_mmu *mmu, phys_addr_t addr, phys_addr_t end);
>   
>   int kvm_handle_guest_abort(struct kvm_vcpu *vcpu);
>   
> diff --git a/arch/arm64/include/asm/kvm_nested.h b/arch/arm64/include/asm/kvm_nested.h
> index f20d272fcd6d..d330b947d48a 100644
> --- a/arch/arm64/include/asm/kvm_nested.h
> +++ b/arch/arm64/include/asm/kvm_nested.h
> @@ -111,6 +111,9 @@ extern int kvm_walk_nested_s2(struct kvm_vcpu *vcpu, phys_addr_t gipa,
>   extern int kvm_s2_handle_perm_fault(struct kvm_vcpu *vcpu,
>   				    struct kvm_s2_trans *trans);
>   extern int kvm_inject_s2_fault(struct kvm_vcpu *vcpu, u64 esr_el2);
> +extern void kvm_nested_s2_wp(struct kvm *kvm);
> +extern void kvm_nested_s2_unmap(struct kvm *kvm);
> +extern void kvm_nested_s2_flush(struct kvm *kvm);
>   int handle_wfx_nested(struct kvm_vcpu *vcpu, bool is_wfe);
>   
>   extern bool forward_smc_trap(struct kvm_vcpu *vcpu);
> diff --git a/arch/arm64/kvm/mmu.c b/arch/arm64/kvm/mmu.c
> index 1e19c59b8235..8144bb9b9ec8 100644
> --- a/arch/arm64/kvm/mmu.c
> +++ b/arch/arm64/kvm/mmu.c
> @@ -245,13 +245,19 @@ void kvm_unmap_stage2_range(struct kvm_s2_mmu *mmu, phys_addr_t start, u64 size)
>   	__unmap_stage2_range(mmu, start, size, true);
>   }
>   
> +void kvm_stage2_flush_range(struct kvm_s2_mmu *mmu,
> +			    phys_addr_t addr, phys_addr_t end)
> +{
> +	stage2_apply_range_resched(mmu, addr, end, kvm_pgtable_stage2_flush);
> +}
> +
>   static void stage2_flush_memslot(struct kvm *kvm,
>   				 struct kvm_memory_slot *memslot)
>   {
>   	phys_addr_t addr = memslot->base_gfn << PAGE_SHIFT;
>   	phys_addr_t end = addr + PAGE_SIZE * memslot->npages;
>   
> -	stage2_apply_range_resched(&kvm->arch.mmu, addr, end, kvm_pgtable_stage2_flush);
> +	kvm_stage2_flush_range(&kvm->arch.mmu, addr, end);
>   }
>   
>   /**
> @@ -274,6 +280,8 @@ static void stage2_flush_vm(struct kvm *kvm)
>   	kvm_for_each_memslot(memslot, bkt, slots)
>   		stage2_flush_memslot(kvm, memslot);
>   
> +	kvm_nested_s2_flush(kvm);
> +
>   	write_unlock(&kvm->mmu_lock);
>   	srcu_read_unlock(&kvm->srcu, idx);
>   }
> @@ -888,6 +896,8 @@ void stage2_unmap_vm(struct kvm *kvm)
>   	kvm_for_each_memslot(memslot, bkt, slots)
>   		stage2_unmap_memslot(kvm, memslot);
>   
> +	kvm_nested_s2_unmap(kvm);
> +
>   	write_unlock(&kvm->mmu_lock);
>   	mmap_read_unlock(current->mm);
>   	srcu_read_unlock(&kvm->srcu, idx);
> @@ -987,12 +997,12 @@ int kvm_phys_addr_ioremap(struct kvm *kvm, phys_addr_t guest_ipa,
>   }
>   
>   /**
> - * stage2_wp_range() - write protect stage2 memory region range
> + * kvm_stage2_wp_range() - write protect stage2 memory region range
>    * @mmu:        The KVM stage-2 MMU pointer
>    * @addr:	Start address of range
>    * @end:	End address of range
>    */
> -static void stage2_wp_range(struct kvm_s2_mmu *mmu, phys_addr_t addr, phys_addr_t end)
> +void kvm_stage2_wp_range(struct kvm_s2_mmu *mmu, phys_addr_t addr, phys_addr_t end)
>   {
>   	stage2_apply_range_resched(mmu, addr, end, kvm_pgtable_stage2_wrprotect);
>   }
> @@ -1023,7 +1033,8 @@ static void kvm_mmu_wp_memory_region(struct kvm *kvm, int slot)
>   	end = (memslot->base_gfn + memslot->npages) << PAGE_SHIFT;
>   
>   	write_lock(&kvm->mmu_lock);
> -	stage2_wp_range(&kvm->arch.mmu, start, end);
> +	kvm_stage2_wp_range(&kvm->arch.mmu, start, end);
> +	kvm_nested_s2_wp(kvm);
>   	write_unlock(&kvm->mmu_lock);
>   	kvm_flush_remote_tlbs(kvm);
>   }
> @@ -1047,7 +1058,7 @@ static void kvm_mmu_write_protect_pt_masked(struct kvm *kvm,
>   	phys_addr_t start = (base_gfn +  __ffs(mask)) << PAGE_SHIFT;
>   	phys_addr_t end = (base_gfn + __fls(mask) + 1) << PAGE_SHIFT;
>   
> -	stage2_wp_range(&kvm->arch.mmu, start, end);
> +	kvm_stage2_wp_range(&kvm->arch.mmu, start, end);
>   }
>   
>   /*
> @@ -1062,6 +1073,7 @@ void kvm_arch_mmu_enable_log_dirty_pt_masked(struct kvm *kvm,
>   		gfn_t gfn_offset, unsigned long mask)
>   {
>   	kvm_mmu_write_protect_pt_masked(kvm, slot, gfn_offset, mask);
> +	kvm_nested_s2_wp(kvm);
>   }
>   
>   static void kvm_send_hwpoison_signal(unsigned long address, short lsb)
> @@ -1720,6 +1732,7 @@ bool kvm_unmap_gfn_range(struct kvm *kvm, struct kvm_gfn_range *range)
>   			     (range->end - range->start) << PAGE_SHIFT,
>   			     range->may_block);
>   
> +	kvm_nested_s2_unmap(kvm);

This kvm_nested_s2_unmap/kvm_unmap_stage2_range is called for every 
active L2 and page table walk-through iterates for long iterations since 
kvm_phys_size(mmu) is pretty big size(atleast 48bits).
What would be the best fix if we want to avoid this unnessary long 
iteration of page table lookup?

>   	return false;
>   }
>   
> @@ -1754,6 +1767,7 @@ bool kvm_set_spte_gfn(struct kvm *kvm, struct kvm_gfn_range *range)
>   			       PAGE_SIZE, __pfn_to_phys(pfn),
>   			       KVM_PGTABLE_PROT_R, NULL, 0);
>   
> +	kvm_nested_s2_unmap(kvm);
>   	return false;
>   }
>   
> @@ -1772,6 +1786,11 @@ bool kvm_age_gfn(struct kvm *kvm, struct kvm_gfn_range *range)
>   					range->start << PAGE_SHIFT);
>   	pte = __pte(kpte);
>   	return pte_valid(pte) && pte_young(pte);
> +
> +	/*
> +	 * TODO: Handle nested_mmu structures here using the reverse mapping in
> +	 * a later version of patch series.
> +	 */
>   }
>   
>   bool kvm_test_age_gfn(struct kvm *kvm, struct kvm_gfn_range *range)
> @@ -2004,6 +2023,7 @@ void kvm_arch_flush_shadow_memslot(struct kvm *kvm,
>   
>   	write_lock(&kvm->mmu_lock);
>   	kvm_unmap_stage2_range(&kvm->arch.mmu, gpa, size);
> +	kvm_nested_s2_unmap(kvm);
>   	write_unlock(&kvm->mmu_lock);
>   }
>   
> diff --git a/arch/arm64/kvm/nested.c b/arch/arm64/kvm/nested.c
> index 73c0be25345a..948ac5b9638c 100644
> --- a/arch/arm64/kvm/nested.c
> +++ b/arch/arm64/kvm/nested.c
> @@ -523,6 +523,45 @@ int kvm_inject_s2_fault(struct kvm_vcpu *vcpu, u64 esr_el2)
>   	return kvm_inject_nested_sync(vcpu, esr_el2);
>   }
>   
> +/* expects kvm->mmu_lock to be held */
> +void kvm_nested_s2_wp(struct kvm *kvm)
> +{
> +	int i;
> +
> +	for (i = 0; i < kvm->arch.nested_mmus_size; i++) {
> +		struct kvm_s2_mmu *mmu = &kvm->arch.nested_mmus[i];
> +
> +		if (kvm_s2_mmu_valid(mmu))
> +			kvm_stage2_wp_range(mmu, 0, kvm_phys_size(mmu));
> +	}
> +}
> +
> +/* expects kvm->mmu_lock to be held */
> +void kvm_nested_s2_unmap(struct kvm *kvm)
> +{
> +	int i;
> +
> +	for (i = 0; i < kvm->arch.nested_mmus_size; i++) {
> +		struct kvm_s2_mmu *mmu = &kvm->arch.nested_mmus[i];
> +
> +		if (kvm_s2_mmu_valid(mmu))
> +			kvm_unmap_stage2_range(mmu, 0, kvm_phys_size(mmu));
> +	}
> +}
> +
> +/* expects kvm->mmu_lock to be held */
> +void kvm_nested_s2_flush(struct kvm *kvm)
> +{
> +	int i;
> +
> +	for (i = 0; i < kvm->arch.nested_mmus_size; i++) {
> +		struct kvm_s2_mmu *mmu = &kvm->arch.nested_mmus[i];
> +
> +		if (kvm_s2_mmu_valid(mmu))
> +			kvm_stage2_flush_range(mmu, 0, kvm_phys_size(mmu));
> +	}
> +}
> +
>   void kvm_arch_flush_shadow_all(struct kvm *kvm)
>   {
>   	int i;

Thanks,
Ganapat