[PATCH v4 02/16] KVM: PPC: Use RCU for arch.spapr_tce_tables

[Alexey Kardashevskiy]

At the moment spapr_tce_tables is not protected against races. This makes
use of RCU-variants of list helpers. As some bits are executed in real
mode, this makes use of just introduced list_for_each_entry_rcu_notrace().

This converts release_spapr_tce_table() to a RCU scheduled handler.

Signed-off-by: Alexey Kardashevskiy <aik at ozlabs.ru>
---
Changes:
* total rework
* kfree() for kvmppc_spapr_tce_table is moved to call_rcu_sched() callback
* used new list_for_each_entry_rcu_notrace
---
 arch/powerpc/include/asm/kvm_host.h |  1 +
 arch/powerpc/kvm/book3s.c           |  2 +-
 arch/powerpc/kvm/book3s_64_vio.c    | 23 +++++++++++++----------
 arch/powerpc/kvm/book3s_64_vio_hv.c |  6 ++++--
 4 files changed, 19 insertions(+), 13 deletions(-)

diff --git a/arch/powerpc/include/asm/kvm_host.h b/arch/powerpc/include/asm/kvm_host.h
index bb66d8b..cd22c31 100644
--- a/arch/powerpc/include/asm/kvm_host.h
+++ b/arch/powerpc/include/asm/kvm_host.h
@@ -180,6 +180,7 @@ struct kvmppc_spapr_tce_table {
 	struct kvm *kvm;
 	u64 liobn;
 	u32 window_size;
+	struct rcu_head rcu;
 	struct page *pages[0];
 };
 
diff --git a/arch/powerpc/kvm/book3s.c b/arch/powerpc/kvm/book3s.c
index c254c27..9e17d19 100644
--- a/arch/powerpc/kvm/book3s.c
+++ b/arch/powerpc/kvm/book3s.c
@@ -886,7 +886,7 @@ int kvmppc_core_init_vm(struct kvm *kvm)
 {
 
 #ifdef CONFIG_PPC64
-	INIT_LIST_HEAD(&kvm->arch.spapr_tce_tables);
+	INIT_LIST_HEAD_RCU(&kvm->arch.spapr_tce_tables);
 	INIT_LIST_HEAD(&kvm->arch.rtas_tokens);
 #endif
 
diff --git a/arch/powerpc/kvm/book3s_64_vio.c b/arch/powerpc/kvm/book3s_64_vio.c
index 54cf9bc..5958f7d 100644
--- a/arch/powerpc/kvm/book3s_64_vio.c
+++ b/arch/powerpc/kvm/book3s_64_vio.c
@@ -45,19 +45,16 @@ static long kvmppc_stt_npages(unsigned long window_size)
 		     * sizeof(u64), PAGE_SIZE) / PAGE_SIZE;
 }
 
-static void release_spapr_tce_table(struct kvmppc_spapr_tce_table *stt)
+static void release_spapr_tce_table(struct rcu_head *head)
 {
-	struct kvm *kvm = stt->kvm;
+	struct kvmppc_spapr_tce_table *stt = container_of(head,
+			struct kvmppc_spapr_tce_table, rcu);
 	int i;
 
-	mutex_lock(&kvm->lock);
-	list_del(&stt->list);
 	for (i = 0; i < kvmppc_stt_npages(stt->window_size); i++)
 		__free_page(stt->pages[i]);
+	kvm_put_kvm(stt->kvm);
 	kfree(stt);
-	mutex_unlock(&kvm->lock);
-
-	kvm_put_kvm(kvm);
 }
 
 static int kvm_spapr_tce_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
@@ -87,8 +84,13 @@ static int kvm_spapr_tce_mmap(struct file *file, struct vm_area_struct *vma)
 static int kvm_spapr_tce_release(struct inode *inode, struct file *filp)
 {
 	struct kvmppc_spapr_tce_table *stt = filp->private_data;
+	struct kvm *kvm = stt->kvm;
+
+	mutex_lock(&kvm->lock);
+	list_del_rcu(&stt->list);
+	call_rcu_sched(&stt->rcu, release_spapr_tce_table);
+	mutex_unlock(&kvm->lock);
 
-	release_spapr_tce_table(stt);
 	return 0;
 }
 
@@ -106,7 +108,8 @@ long kvm_vm_ioctl_create_spapr_tce(struct kvm *kvm,
 	int i;
 
 	/* Check this LIOBN hasn't been previously allocated */
-	list_for_each_entry(stt, &kvm->arch.spapr_tce_tables, list) {
+	list_for_each_entry_rcu_notrace(stt, &kvm->arch.spapr_tce_tables,
+			list) {
 		if (stt->liobn == args->liobn)
 			return -EBUSY;
 	}
@@ -131,7 +134,7 @@ long kvm_vm_ioctl_create_spapr_tce(struct kvm *kvm,
 	kvm_get_kvm(kvm);
 
 	mutex_lock(&kvm->lock);
-	list_add(&stt->list, &kvm->arch.spapr_tce_tables);
+	list_add_rcu(&stt->list, &kvm->arch.spapr_tce_tables);
 
 	mutex_unlock(&kvm->lock);
 
diff --git a/arch/powerpc/kvm/book3s_64_vio_hv.c b/arch/powerpc/kvm/book3s_64_vio_hv.c
index 89e96b3..b1914d9 100644
--- a/arch/powerpc/kvm/book3s_64_vio_hv.c
+++ b/arch/powerpc/kvm/book3s_64_vio_hv.c
@@ -50,7 +50,8 @@ long kvmppc_h_put_tce(struct kvm_vcpu *vcpu, unsigned long liobn,
 	/* udbg_printf("H_PUT_TCE(): liobn=0x%lx ioba=0x%lx, tce=0x%lx\n", */
 	/* 	    liobn, ioba, tce); */
 
-	list_for_each_entry(stt, &kvm->arch.spapr_tce_tables, list) {
+	list_for_each_entry_rcu_notrace(stt, &kvm->arch.spapr_tce_tables,
+			list) {
 		if (stt->liobn == liobn) {
 			unsigned long idx = ioba >> SPAPR_TCE_SHIFT;
 			struct page *page;
@@ -82,7 +83,8 @@ long kvmppc_h_get_tce(struct kvm_vcpu *vcpu, unsigned long liobn,
 	struct kvm *kvm = vcpu->kvm;
 	struct kvmppc_spapr_tce_table *stt;
 
-	list_for_each_entry(stt, &kvm->arch.spapr_tce_tables, list) {
+	list_for_each_entry_rcu_notrace(stt, &kvm->arch.spapr_tce_tables,
+			list) {
 		if (stt->liobn == liobn) {
 			unsigned long idx = ioba >> SPAPR_TCE_SHIFT;
 			struct page *page;
-- 
2.0.0