diff mbox series

[v2,13/16] KVM: arm64: Run clear-dirty-log under MMU read lock

Message ID 20230602160914.4011728-14-vipinsh@google.com (mailing list archive)
State New, archived
Headers show
Series Use MMU read lock for clear-dirty-log | expand

Commit Message

Vipin Sharma June 2, 2023, 4:09 p.m. UTC
Take MMU read lock for clearing dirty logs and use shared page table
walker.

Dirty logs are currently cleared using MMU write locks. This
means vCPUs page faults, which takes MMU read lock,  will
be blocked while dirty logs are being cleared. This causes guest
degradation and especially noticeable on VMs with lot of vCPUs.

Taking MMU read lock will allow vCPUs to execute parallelly and reduces
the impact on vCPUs performance.

Signed-off-by: Vipin Sharma <vipinsh@google.com>
---
 arch/arm64/kvm/mmu.c | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)
diff mbox series

Patch

diff --git a/arch/arm64/kvm/mmu.c b/arch/arm64/kvm/mmu.c
index 356dc4131023..7c966f6f1a41 100644
--- a/arch/arm64/kvm/mmu.c
+++ b/arch/arm64/kvm/mmu.c
@@ -74,8 +74,12 @@  static int stage2_apply_range(struct kvm_s2_mmu *mmu, phys_addr_t addr,
 		if (ret)
 			break;
 
-		if (resched && next != end)
-			cond_resched_rwlock_write(&kvm->mmu_lock);
+		if (resched && next != end) {
+			if (flags & KVM_PGTABLE_WALK_SHARED)
+				cond_resched_rwlock_read(&kvm->mmu_lock);
+			else
+				cond_resched_rwlock_write(&kvm->mmu_lock);
+		}
 	} while (addr = next, addr != end);
 
 	return ret;
@@ -1131,11 +1135,11 @@  void kvm_arch_mmu_enable_log_dirty_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;
 
-	write_lock(&kvm->mmu_lock);
-	lockdep_assert_held_write(&kvm->mmu_lock);
-
-	stage2_wp_range(&kvm->arch.mmu, start, end, 0);
+	read_lock(&kvm->mmu_lock);
+	stage2_wp_range(&kvm->arch.mmu, start, end, KVM_PGTABLE_WALK_SHARED);
+	read_unlock(&kvm->mmu_lock);
 
+	write_lock(&kvm->mmu_lock);
 	/*
 	 * Eager-splitting is done when manual-protect is set.  We
 	 * also check for initially-all-set because we can avoid