Commit fbb158cb authored by Sean Christopherson's avatar Sean Christopherson Committed by Paolo Bonzini

KVM: x86/mmu: Revert "Revert "KVM: MMU: zap pages in batch""

Now that the fast invalidate mechanism has been reintroduced, restore
the performance tweaks for fast invalidation that existed prior to its
removal.

Paraphrashing the original changelog:

  Zap at least 10 shadow pages before releasing mmu_lock to reduce the
  overhead associated with re-acquiring the lock.

  Note: "10" is an arbitrary number, speculated to be high enough so
  that a vCPU isn't stuck zapping obsolete pages for an extended period,
  but small enough so that other vCPUs aren't starved waiting for
  mmu_lock.

This reverts commit 43d2b14b.
Signed-off-by: default avatarSean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: default avatarPaolo Bonzini <pbonzini@redhat.com>
parent 14a3c4f4
...@@ -5671,12 +5671,12 @@ int kvm_mmu_create(struct kvm_vcpu *vcpu) ...@@ -5671,12 +5671,12 @@ int kvm_mmu_create(struct kvm_vcpu *vcpu)
return ret; return ret;
} }
#define BATCH_ZAP_PAGES 10
static void kvm_zap_obsolete_pages(struct kvm *kvm) static void kvm_zap_obsolete_pages(struct kvm *kvm)
{ {
struct kvm_mmu_page *sp, *node; struct kvm_mmu_page *sp, *node;
LIST_HEAD(invalid_list); LIST_HEAD(invalid_list);
int ign; int nr_zapped, batch = 0;
restart: restart:
list_for_each_entry_safe_reverse(sp, node, list_for_each_entry_safe_reverse(sp, node,
...@@ -5689,28 +5689,6 @@ static void kvm_zap_obsolete_pages(struct kvm *kvm) ...@@ -5689,28 +5689,6 @@ static void kvm_zap_obsolete_pages(struct kvm *kvm)
break; break;
/* /*
* Do not repeatedly zap a root page to avoid unnecessary
* KVM_REQ_MMU_RELOAD, otherwise we may not be able to
* progress:
* vcpu 0 vcpu 1
* call vcpu_enter_guest():
* 1): handle KVM_REQ_MMU_RELOAD
* and require mmu-lock to
* load mmu
* repeat:
* 1): zap root page and
* send KVM_REQ_MMU_RELOAD
*
* 2): if (cond_resched_lock(mmu-lock))
*
* 2): hold mmu-lock and load mmu
*
* 3): see KVM_REQ_MMU_RELOAD bit
* on vcpu->requests is set
* then return 1 to call
* vcpu_enter_guest() again.
* goto repeat;
*
* Since we are reversely walking the list and the invalid * Since we are reversely walking the list and the invalid
* list will be moved to the head, skip the invalid page * list will be moved to the head, skip the invalid page
* can help us to avoid the infinity list walking. * can help us to avoid the infinity list walking.
...@@ -5718,14 +5696,19 @@ static void kvm_zap_obsolete_pages(struct kvm *kvm) ...@@ -5718,14 +5696,19 @@ static void kvm_zap_obsolete_pages(struct kvm *kvm)
if (sp->role.invalid) if (sp->role.invalid)
continue; continue;
if (need_resched() || spin_needbreak(&kvm->mmu_lock)) { if (batch >= BATCH_ZAP_PAGES &&
(need_resched() || spin_needbreak(&kvm->mmu_lock))) {
batch = 0;
kvm_mmu_commit_zap_page(kvm, &invalid_list); kvm_mmu_commit_zap_page(kvm, &invalid_list);
cond_resched_lock(&kvm->mmu_lock); cond_resched_lock(&kvm->mmu_lock);
goto restart; goto restart;
} }
if (__kvm_mmu_prepare_zap_page(kvm, sp, &invalid_list, &ign)) if (__kvm_mmu_prepare_zap_page(kvm, sp, &invalid_list,
&nr_zapped)) {
batch += nr_zapped;
goto restart; goto restart;
}
} }
kvm_mmu_commit_zap_page(kvm, &invalid_list); kvm_mmu_commit_zap_page(kvm, &invalid_list);
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment