[12/14] Revert "KVM: x86: use the fast way to invalidate all pages"
diff mbox series

Message ID 20190110180706.24974-13-sean.j.christopherson@intel.com
State New
Headers show
  • KVM: x86/mmu: Remove fast invalidate mechanism
Related show

Commit Message

Sean Christopherson Jan. 10, 2019, 6:07 p.m. UTC
Revert to a slow kvm_mmu_zap_all() for kvm_arch_flush_shadow_all().
Flushing all shadow entries is only done during VM teardown, i.e.
kvm_arch_flush_shadow_all() is only called when the associated MM struct
is being released or when the VM instance is being freed.

In normal operation there aren't any active vCPUs to defer to, and in an
unexpected teardown KVM deferring to running vCPUs is not desirable.

An argument can be made that KVM should still schedule voluntarily
during VM teardown (since it's a common path) to play nice with the rest
of the kernel, but during teardown that can be done without the fast
invalidate mechanism, e.g. marking the VM as dead to prevent MMU

This reverts commit 6ca18b6950f8dee29361722f28f69847724b276f.

Cc: Xiao Guangrong <guangrong.xiao@gmail.com>
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
 arch/x86/kvm/mmu.c | 15 +++++++++++++++
 arch/x86/kvm/x86.c |  2 +-
 2 files changed, 16 insertions(+), 1 deletion(-)

diff mbox series

diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c
index ce5d5ec99fa5..291356af4a00 100644
--- a/arch/x86/kvm/mmu.c
+++ b/arch/x86/kvm/mmu.c
@@ -5836,6 +5836,21 @@  void kvm_mmu_slot_set_dirty(struct kvm *kvm,
+void kvm_mmu_zap_all(struct kvm *kvm)
+	struct kvm_mmu_page *sp, *node;
+	LIST_HEAD(invalid_list);
+	spin_lock(&kvm->mmu_lock);
+	list_for_each_entry_safe(sp, node, &kvm->arch.active_mmu_pages, link)
+		if (kvm_mmu_prepare_zap_page(kvm, sp, &invalid_list))
+			goto restart;
+	kvm_mmu_commit_zap_page(kvm, &invalid_list);
+	spin_unlock(&kvm->mmu_lock);
 static void kvm_zap_obsolete_pages(struct kvm *kvm)
 	struct kvm_mmu_page *sp, *node;
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index bd903779610e..5d7def8fabe2 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -9450,7 +9450,7 @@  void kvm_arch_commit_memory_region(struct kvm *kvm,
 void kvm_arch_flush_shadow_all(struct kvm *kvm)
-	kvm_mmu_invalidate_zap_all_pages(kvm);
+	kvm_mmu_zap_all(kvm);
 void kvm_arch_flush_shadow_memslot(struct kvm *kvm,