Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.33 as permitted sender) client-ip=23.128.96.33;
From:   Paolo Bonzini <pbonzini@redhat.com>
To:     linux-kernel@vger.kernel.org, kvm@vger.kernel.org
Subject: [PATCH 3/3] KVM: x86/mmu: always take tdp_mmu_pages_lock
Date:   Thu, 28 Sep 2023 12:29:59 -0400
Message-Id: <20230928162959.1514661-4-pbonzini@redhat.com>
In-Reply-To: <20230928162959.1514661-1-pbonzini@redhat.com>
References: <20230928162959.1514661-1-pbonzini@redhat.com>
MIME-Version: 1.0
Content-Type: text/plain
Content-Transfer-Encoding: 8bit
Precedence: bulk

It is cheap to take tdp_mmu_pages_lock in all write-side critical sections.
We already do it all the time when zapping with read_lock(), so it is not
a problem to do it from the kvm_tdp_mmu_zap_all() path (aka
kvm_arch_flush_shadow_all(), aka VM destruction and MMU notifier release).

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 Documentation/virt/kvm/locking.rst |  6 ++----
 arch/x86/kvm/mmu/tdp_mmu.c         | 17 ++++++-----------
 2 files changed, 8 insertions(+), 15 deletions(-)

diff --git a/Documentation/virt/kvm/locking.rst b/Documentation/virt/kvm/locking.rst
index 3a034db5e55f..381eb0e7d947 100644
--- a/Documentation/virt/kvm/locking.rst
+++ b/Documentation/virt/kvm/locking.rst
@@ -43,10 +43,8 @@ On x86:
 
 - vcpu->mutex is taken outside kvm->arch.hyperv.hv_lock and kvm->arch.xen.xen_lock
 
-- kvm->arch.mmu_lock is an rwlock.  kvm->arch.tdp_mmu_pages_lock and
-  kvm->arch.mmu_unsync_pages_lock are taken inside kvm->arch.mmu_lock, and
-  cannot be taken without already holding kvm->arch.mmu_lock (typically with
-  ``read_lock`` for the TDP MMU, thus the need for additional spinlocks).
+- kvm->arch.mmu_lock is an rwlock and is taken outside
+  kvm->arch.tdp_mmu_pages_lock and kvm->arch.mmu_unsync_pages_lock
 
 Everything else is a leaf: no other lock is taken inside the critical
 sections.
diff --git a/arch/x86/kvm/mmu/tdp_mmu.c b/arch/x86/kvm/mmu/tdp_mmu.c
index b9abfa78808a..f61bc842067f 100644
--- a/arch/x86/kvm/mmu/tdp_mmu.c
+++ b/arch/x86/kvm/mmu/tdp_mmu.c
@@ -285,24 +285,19 @@ static void tdp_unaccount_mmu_page(struct kvm *kvm, struct kvm_mmu_page *sp)
  *	    the MMU lock and the operation must synchronize with other
  *	    threads that might be adding or removing pages.
  */
-static void tdp_mmu_unlink_sp(struct kvm *kvm, struct kvm_mmu_page *sp,
-			      bool shared)
+static void tdp_mmu_unlink_sp(struct kvm *kvm, struct kvm_mmu_page *sp)
 {
+	lockdep_assert_held(&kvm->mmu_lock);
+
 	tdp_unaccount_mmu_page(kvm, sp);
 
 	if (!sp->nx_huge_page_disallowed)
 		return;
 
-	if (shared)
-		spin_lock(&kvm->arch.tdp_mmu_pages_lock);
-	else
-		lockdep_assert_held_write(&kvm->mmu_lock);
-
+	spin_lock(&kvm->arch.tdp_mmu_pages_lock);
 	sp->nx_huge_page_disallowed = false;
 	untrack_possible_nx_huge_page(kvm, sp);
-
-	if (shared)
-		spin_unlock(&kvm->arch.tdp_mmu_pages_lock);
+	spin_unlock(&kvm->arch.tdp_mmu_pages_lock);
 }
 
 /**
@@ -331,7 +326,7 @@ static void handle_removed_pt(struct kvm *kvm, tdp_ptep_t pt, bool shared)
 
 	trace_kvm_mmu_prepare_zap_page(sp);
 
-	tdp_mmu_unlink_sp(kvm, sp, shared);
+	tdp_mmu_unlink_sp(kvm, sp);
 
 	for (i = 0; i < SPTE_ENT_PER_PAGE; i++) {
 		tdp_ptep_t sptep = pt + i;
-- 
2.39.1