[Patch v2 2/4] KVM:SVM: Implement pin_page support

[eric van tassell]

Improve SEV guest startup time from O(n) to a constant by deferring
guest page pinning until the pages are used to satisfy nested page faults.

Implement the code to do the pinning (sev_get_page) and the notifier
sev_pin_page().

Track the pinned pages with xarray so they can be released during guest
termination.

Co-developed-by: Brijesh Singh <brijesh.singh@xxxxxxx>
Signed-off-by: eric van tassell <Eric.VanTassell@xxxxxxx>
---
 arch/x86/kvm/svm/sev.c | 68 ++++++++++++++++++++++++++++++++++++++++++
 arch/x86/kvm/svm/svm.c |  2 ++
 arch/x86/kvm/svm/svm.h |  3 ++
 3 files changed, 73 insertions(+)

diff --git a/arch/x86/kvm/svm/sev.c b/arch/x86/kvm/svm/sev.c
index f7f1f4ecf08e..8d56d1afb33e 100644
--- a/arch/x86/kvm/svm/sev.c
+++ b/arch/x86/kvm/svm/sev.c
@@ -184,6 +184,8 @@ static int sev_guest_init(struct kvm *kvm, struct kvm_sev_cmd *argp)
 	sev->asid = asid;
 	INIT_LIST_HEAD(&sev->regions_list);
 
+	xa_init(&sev->pages_xarray);
+
 	return 0;
 
 e_free:
@@ -415,6 +417,43 @@ static unsigned long get_num_contig_pages(unsigned long idx,
 	return pages;
 }
 
+static int sev_get_page(struct kvm *kvm, gfn_t gfn, kvm_pfn_t pfn)
+{
+	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
+	struct xarray *xa = &sev->pages_xarray;
+	struct page *page = pfn_to_page(pfn);
+	int ret;
+
+	/* store page at index = gfn */
+	ret = xa_insert(xa, gfn, page, GFP_ATOMIC);
+	if (ret == -EBUSY) {
+		/*
+		 * If xa_insert returned -EBUSY, the  gfn was already associated
+		 * with a struct page *.
+		 */
+		struct page *cur_page;
+
+		cur_page = xa_load(xa, gfn);
+		/* If cur_page == page, no change is needed, so return 0 */
+		if (cur_page == page)
+			return 0;
+
+		/* Release the page that was stored at index = gfn */
+		put_page(cur_page);
+
+		/* Return result of attempting to store page at index = gfn */
+		ret = xa_err(xa_store(xa, gfn, page, GFP_ATOMIC));
+		WARN_ON(ret != 0);
+	}
+
+	if (ret)
+		return ret;
+
+	get_page(page);
+
+	return 0;
+}
+
 static int sev_launch_update_data(struct kvm *kvm, struct kvm_sev_cmd *argp)
 {
 	unsigned long vaddr, vaddr_end, next_vaddr, npages, pages, size, i;
@@ -1085,6 +1124,8 @@ void sev_vm_destroy(struct kvm *kvm)
 	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
 	struct list_head *head = &sev->regions_list;
 	struct list_head *pos, *q;
+	XA_STATE(xas, &sev->pages_xarray, 0);
+	struct page *xa_page;
 
 	if (!sev_guest(kvm))
 		return;
@@ -1109,6 +1150,12 @@ void sev_vm_destroy(struct kvm *kvm)
 		}
 	}
 
+	/* Release each pinned page that SEV tracked in sev->pages_xarray. */
+	xas_for_each(&xas, xa_page, ULONG_MAX) {
+		put_page(xa_page);
+	}
+	xa_destroy(&sev->pages_xarray);
+
 	mutex_unlock(&kvm->lock);
 
 	sev_unbind_asid(kvm, sev->handle);
@@ -1193,3 +1240,24 @@ void pre_sev_run(struct vcpu_svm *svm, int cpu)
 	svm->vmcb->control.tlb_ctl = TLB_CONTROL_FLUSH_ASID;
 	vmcb_mark_dirty(svm->vmcb, VMCB_ASID);
 }
+
+int sev_pin_page(struct kvm_vcpu *vcpu, gfn_t gfn, kvm_pfn_t pfn,
+		 int level, u64 *spte)
+{
+	int rc;
+
+	if (!sev_guest(vcpu->kvm))
+		return 0;
+
+	rc = sev_get_page(vcpu->kvm, gfn, pfn);
+	if (rc)
+		return rc;
+
+	/*
+	 * Flush any cached lines of the page being added since "ownership" of
+	 * it will be transferred from the host to an encrypted guest.
+	 */
+	clflush_cache_range(__va(pfn << PAGE_SHIFT), page_level_size(level));
+
+	return 0;
+}
diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c
index 535ad311ad02..adb308631416 100644
--- a/arch/x86/kvm/svm/svm.c
+++ b/arch/x86/kvm/svm/svm.c
@@ -4130,6 +4130,8 @@ static struct kvm_x86_ops svm_x86_ops __initdata = {
 	.need_emulation_on_page_fault = svm_need_emulation_on_page_fault,
 
 	.apic_init_signal_blocked = svm_apic_init_signal_blocked,
+
+	.pin_page = sev_pin_page,
 };
 
 static struct kvm_x86_init_ops svm_init_ops __initdata = {
diff --git a/arch/x86/kvm/svm/svm.h b/arch/x86/kvm/svm/svm.h
index 121b198b51e9..278c46bc52aa 100644
--- a/arch/x86/kvm/svm/svm.h
+++ b/arch/x86/kvm/svm/svm.h
@@ -65,6 +65,7 @@ struct kvm_sev_info {
 	int fd;			/* SEV device fd */
 	unsigned long pages_locked; /* Number of pages locked */
 	struct list_head regions_list;  /* List of registered regions */
+	struct xarray pages_xarray; /* List of PFN locked */
 };
 
 struct kvm_svm {
@@ -488,5 +489,7 @@ int svm_unregister_enc_region(struct kvm *kvm,
 void pre_sev_run(struct vcpu_svm *svm, int cpu);
 int __init sev_hardware_setup(void);
 void sev_hardware_teardown(void);
+int sev_pin_page(struct kvm_vcpu *vcpu, gfn_t gfn, kvm_pfn_t pfn,
+		 int level, u64 *spte);
 
 #endif
-- 
2.17.1