[PATCH] KVM: Use precise range-based flush in mmu_notifier hooks when possible

[Sean Christopherson]

Do arch-specific range-based TLB flushes (if they're supported) when
flushing in response to mmu_notifier events, as a single range-based flush
is almost always more performant.  This is especially true in the case of
mmu_notifier events, as the majority of events that hit a running VM
operate on a relatively small range of memory.

Cc: Marc Zyngier <maz@xxxxxxxxxx>
Cc: Will Deacon <will@xxxxxxxxxx>
Signed-off-by: Sean Christopherson <seanjc@xxxxxxxxxx>
---

This is *very* lightly tested, a thumbs up from the ARM world would be much
appreciated.

 virt/kvm/kvm_main.c | 15 ++++++++++++++-
 1 file changed, 14 insertions(+), 1 deletion(-)

diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index d0788d0a72cc..46bb95d58d53 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -599,6 +599,7 @@ static __always_inline kvm_mn_ret_t __kvm_handle_hva_range(struct kvm *kvm,
 	struct kvm_gfn_range gfn_range;
 	struct kvm_memory_slot *slot;
 	struct kvm_memslots *slots;
+	bool need_flush = false;
 	int i, idx;
 
 	if (WARN_ON_ONCE(range->end <= range->start))
@@ -651,10 +652,22 @@ static __always_inline kvm_mn_ret_t __kvm_handle_hva_range(struct kvm *kvm,
 					goto mmu_unlock;
 			}
 			r.ret |= range->handler(kvm, &gfn_range);
+
+			/*
+			 * Use a precise gfn-based TLB flush when possible, as
+			 * most mmu_notifier events affect a small-ish range.
+			 * Fall back to a full TLB flush if the gfn-based flush
+			 * fails, and don't bother trying the gfn-based flush
+			 * if a full flush is already pending.
+			 */
+			if (range->flush_on_ret && !need_flush && r.ret &&
+			    kvm_arch_flush_remote_tlbs_range(kvm, gfn_range.start,
+							     gfn_range.end - gfn_range.start))
+				need_flush = true;
 		}
 	}
 
-	if (range->flush_on_ret && r.ret)
+	if (need_flush)
 		kvm_flush_remote_tlbs(kvm);
 
 mmu_unlock:

base-commit: 332d2c1d713e232e163386c35a3ba0c1b90df83f
-- 
2.46.0.rc2.264.g509ed76dc8-goog