kvm: x86: mmu: Recovery of shattered NX large pages

commit 1aa9b9572b10529c2e64e2b8f44025d86e124308 upstream.

The page table pages corresponding to broken down large pages are zapped in
FIFO order, so that the large page can potentially be recovered, if it is
not longer being used for execution.  This removes the performance penalty
for walking deeper EPT page tables.

By default, one large page will last about one hour once the guest
reaches a steady state.

Signed-off-by: Junaid Shahid <junaids@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
[bwh: Backported to 4.9:
 - Update another error path in kvm_create_vm() to use out_err_no_mmu_notifier
 - Adjust filename, context]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

1 files changed, 129 insertions, 0 deletions

diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c
index 5fbc8677c6f3..f0f180158c26 100644
--- a/arch/x86/kvm/mmu.c
+++ b/arch/x86/kvm/mmu.c
@@ -37,6 +37,7 @@
 #include <linux/srcu.h>
 #include <linux/slab.h>
 #include <linux/uaccess.h>
+#include <linux/kthread.h>
 
 #include <asm/page.h>
 #include <asm/cmpxchg.h>
@@ -47,16 +48,26 @@
 extern bool itlb_multihit_kvm_mitigation;
 
 static int __read_mostly nx_huge_pages = -1;
+static uint __read_mostly nx_huge_pages_recovery_ratio = 60;
 
 static int set_nx_huge_pages(const char *val, const struct kernel_param *kp);
+static int set_nx_huge_pages_recovery_ratio(const char *val, const struct kernel_param *kp);
 
 static struct kernel_param_ops nx_huge_pages_ops = {
 	.set = set_nx_huge_pages,
 	.get = param_get_bool,
 };
 
+static struct kernel_param_ops nx_huge_pages_recovery_ratio_ops = {
+	.set = set_nx_huge_pages_recovery_ratio,
+	.get = param_get_uint,
+};
+
 module_param_cb(nx_huge_pages, &nx_huge_pages_ops, &nx_huge_pages, 0644);
 __MODULE_PARM_TYPE(nx_huge_pages, "bool");
+module_param_cb(nx_huge_pages_recovery_ratio, &nx_huge_pages_recovery_ratio_ops,
+		&nx_huge_pages_recovery_ratio, 0644);
+__MODULE_PARM_TYPE(nx_huge_pages_recovery_ratio, "uint");
 
 /*
  * When setting this variable to true it enables Two-Dimensional-Paging
@@ -880,6 +891,8 @@ static void account_huge_nx_page(struct kvm *kvm, struct kvm_mmu_page *sp)
 		return;
 
 	++kvm->stat.nx_lpage_splits;
+	list_add_tail(&sp->lpage_disallowed_link,
+		      &kvm->arch.lpage_disallowed_mmu_pages);
 	sp->lpage_disallowed = true;
 }
 
@@ -904,6 +917,7 @@ static void unaccount_huge_nx_page(struct kvm *kvm, struct kvm_mmu_page *sp)
 {
 	--kvm->stat.nx_lpage_splits;
 	sp->lpage_disallowed = false;
+	list_del(&sp->lpage_disallowed_link);
 }
 
 static bool __mmu_gfn_lpage_is_disallowed(gfn_t gfn, int level,
@@ -5172,6 +5186,8 @@ static int set_nx_huge_pages(const char *val, const struct kernel_param *kp)
 			idx = srcu_read_lock(&kvm->srcu);
 			kvm_mmu_invalidate_zap_all_pages(kvm);
 			srcu_read_unlock(&kvm->srcu, idx);
+
+			wake_up_process(kvm->arch.nx_lpage_recovery_thread);
 		}
 		mutex_unlock(&kvm_lock);
 	}
@@ -5247,3 +5263,116 @@ void kvm_mmu_module_exit(void)
 	unregister_shrinker(&mmu_shrinker);
 	mmu_audit_disable();
 }
+
+static int set_nx_huge_pages_recovery_ratio(const char *val, const struct kernel_param *kp)
+{
+	unsigned int old_val;
+	int err;
+
+	old_val = nx_huge_pages_recovery_ratio;
+	err = param_set_uint(val, kp);
+	if (err)
+		return err;
+
+	if (READ_ONCE(nx_huge_pages) &&
+	    !old_val && nx_huge_pages_recovery_ratio) {
+		struct kvm *kvm;
+
+		mutex_lock(&kvm_lock);
+
+		list_for_each_entry(kvm, &vm_list, vm_list)
+			wake_up_process(kvm->arch.nx_lpage_recovery_thread);
+
+		mutex_unlock(&kvm_lock);
+	}
+
+	return err;
+}
+
+static void kvm_recover_nx_lpages(struct kvm *kvm)
+{
+	int rcu_idx;
+	struct kvm_mmu_page *sp;
+	unsigned int ratio;
+	LIST_HEAD(invalid_list);
+	ulong to_zap;
+
+	rcu_idx = srcu_read_lock(&kvm->srcu);
+	spin_lock(&kvm->mmu_lock);
+
+	ratio = READ_ONCE(nx_huge_pages_recovery_ratio);
+	to_zap = ratio ? DIV_ROUND_UP(kvm->stat.nx_lpage_splits, ratio) : 0;
+	while (to_zap && !list_empty(&kvm->arch.lpage_disallowed_mmu_pages)) {
+		/*
+		 * We use a separate list instead of just using active_mmu_pages
+		 * because the number of lpage_disallowed pages is expected to
+		 * be relatively small compared to the total.
+		 */
+		sp = list_first_entry(&kvm->arch.lpage_disallowed_mmu_pages,
+				      struct kvm_mmu_page,
+				      lpage_disallowed_link);
+		WARN_ON_ONCE(!sp->lpage_disallowed);
+		kvm_mmu_prepare_zap_page(kvm, sp, &invalid_list);
+		WARN_ON_ONCE(sp->lpage_disallowed);
+
+		if (!--to_zap || need_resched() || spin_needbreak(&kvm->mmu_lock)) {
+			kvm_mmu_commit_zap_page(kvm, &invalid_list);
+			if (to_zap)
+				cond_resched_lock(&kvm->mmu_lock);
+		}
+	}
+
+	spin_unlock(&kvm->mmu_lock);
+	srcu_read_unlock(&kvm->srcu, rcu_idx);
+}
+
+static long get_nx_lpage_recovery_timeout(u64 start_time)
+{
+	return READ_ONCE(nx_huge_pages) && READ_ONCE(nx_huge_pages_recovery_ratio)
+		? start_time + 60 * HZ - get_jiffies_64()
+		: MAX_SCHEDULE_TIMEOUT;
+}
+
+static int kvm_nx_lpage_recovery_worker(struct kvm *kvm, uintptr_t data)
+{
+	u64 start_time;
+	long remaining_time;
+
+	while (true) {
+		start_time = get_jiffies_64();
+		remaining_time = get_nx_lpage_recovery_timeout(start_time);
+
+		set_current_state(TASK_INTERRUPTIBLE);
+		while (!kthread_should_stop() && remaining_time > 0) {
+			schedule_timeout(remaining_time);
+			remaining_time = get_nx_lpage_recovery_timeout(start_time);
+			set_current_state(TASK_INTERRUPTIBLE);
+		}
+
+		set_current_state(TASK_RUNNING);
+
+		if (kthread_should_stop())
+			return 0;
+
+		kvm_recover_nx_lpages(kvm);
+	}
+}
+
+int kvm_mmu_post_init_vm(struct kvm *kvm)
+{
+	int err;
+
+	err = kvm_vm_create_worker_thread(kvm, kvm_nx_lpage_recovery_worker, 0,
+					  "kvm-nx-lpage-recovery",
+					  &kvm->arch.nx_lpage_recovery_thread);
+	if (!err)
+		kthread_unpark(kvm->arch.nx_lpage_recovery_thread);
+
+	return err;
+}
+
+void kvm_mmu_pre_destroy_vm(struct kvm *kvm)
+{
+	if (kvm->arch.nx_lpage_recovery_thread)
+		kthread_stop(kvm->arch.nx_lpage_recovery_thread);
+}