Commit ebdb292d authored Jun 23, 2020 by Sean Christopherson Committed by Paolo Bonzini Jul 08, 2020

KVM: x86/mmu: Batch zap MMU pages when shrinking the slab

Use the recently introduced kvm_mmu_zap_oldest_mmu_pages() to batch zap
MMU pages when shrinking a slab. This fixes a long standing issue where
KVM's shrinker implementation is completely ineffective due to zapping
only a single page. E.g. without batch zapping, forcing a scan via
drop_caches basically has no impact on a VM with ~2k shadow pages. With
batch zapping, the number of shadow pages can be reduced to a few
hundred pages in one or two runs of drop_caches.

Note, if the default batch size (currently 128) is problematic, e.g.
zapping 128 pages holds mmu_lock for too long, KVM can bound the batch
size by setting @batch in mmu_shrinker.

Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Message-Id: <20200623193542.7554-4-sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

parent 6b82ef2c

arch/x86/kvm/mmu/mmu.c

+1 −16

Original line number	Diff line number	Diff line
		@@ -2816,19 +2816,6 @@ static void kvm_mmu_commit_zap_page(struct kvm *kvm,
		}
		}

		static bool prepare_zap_oldest_mmu_page(struct kvm *kvm,
		struct list_head *invalid_list)
		{
		struct kvm_mmu_page *sp;

		if (list_empty(&kvm->arch.active_mmu_pages))
		return false;

		sp = list_last_entry(&kvm->arch.active_mmu_pages,
		struct kvm_mmu_page, link);
		return kvm_mmu_prepare_zap_page(kvm, sp, invalid_list);
		}

		static unsigned long kvm_mmu_zap_oldest_mmu_pages(struct kvm *kvm,
		unsigned long nr_to_zap)
		{
		@@ -6116,9 +6103,7 @@ mmu_shrink_scan(struct shrinker shrink, struct shrink_control sc)
		goto unlock;
		}

		if (prepare_zap_oldest_mmu_page(kvm, &invalid_list))
		freed++;
		kvm_mmu_commit_zap_page(kvm, &invalid_list);
		freed = kvm_mmu_zap_oldest_mmu_pages(kvm, sc->nr_to_scan);

		unlock:
		spin_unlock(&kvm->mmu_lock);

Admin message