s390/mm: add pte invalidation notifier for kvm

Add a notifier for kvm to get control before a page table entry is
invalidated. The notifier is only called for ptes of an address space
with pgstes that have been explicitly marked to require notification.
Kvm will use this to get control before prefix pages of virtual CPU
are unmapped.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

1 files changed, 117 insertions, 4 deletions

diff --git a/arch/s390/mm/pgtable.c b/arch/s390/mm/pgtable.c
index bd954e9..7805ddc 100644
--- a/arch/s390/mm/pgtable.c
+++ b/arch/s390/mm/pgtable.c
@@ -454,9 +454,8 @@ unsigned long gmap_translate(unsigned long address, struct gmap *gmap)
 }
 EXPORT_SYMBOL_GPL(gmap_translate);
 
-static int gmap_connect_pgtable(unsigned long segment,
-				unsigned long *segment_ptr,
-				struct gmap *gmap)
+static int gmap_connect_pgtable(unsigned long address, unsigned long segment,
+				unsigned long *segment_ptr, struct gmap *gmap)
 {
 	unsigned long vmaddr;
 	struct vm_area_struct *vma;
@@ -491,7 +490,9 @@ static int gmap_connect_pgtable(unsigned long segment,
 	/* Link gmap segment table entry location to page table. */
 	page = pmd_page(*pmd);
 	mp = (struct gmap_pgtable *) page->index;
+	rmap->gmap = gmap;
 	rmap->entry = segment_ptr;
+	rmap->vmaddr = address;
 	spin_lock(&mm->page_table_lock);
 	if (*segment_ptr == segment) {
 		list_add(&rmap->list, &mp->mapper);
@@ -553,7 +554,7 @@ unsigned long __gmap_fault(unsigned long address, struct gmap *gmap)
 		if (!(segment & _SEGMENT_ENTRY_RO))
 			/* Nothing mapped in the gmap address space. */
 			break;
-		rc = gmap_connect_pgtable(segment, segment_ptr, gmap);
+		rc = gmap_connect_pgtable(address, segment, segment_ptr, gmap);
 		if (rc)
 			return rc;
 	}
@@ -619,6 +620,118 @@ void gmap_discard(unsigned long from, unsigned long to, struct gmap *gmap)
 }
 EXPORT_SYMBOL_GPL(gmap_discard);
 
+static LIST_HEAD(gmap_notifier_list);
+static DEFINE_SPINLOCK(gmap_notifier_lock);
+
+/**
+ * gmap_register_ipte_notifier - register a pte invalidation callback
+ * @nb: pointer to the gmap notifier block
+ */
+void gmap_register_ipte_notifier(struct gmap_notifier *nb)
+{
+	spin_lock(&gmap_notifier_lock);
+	list_add(&nb->list, &gmap_notifier_list);
+	spin_unlock(&gmap_notifier_lock);
+}
+EXPORT_SYMBOL_GPL(gmap_register_ipte_notifier);
+
+/**
+ * gmap_unregister_ipte_notifier - remove a pte invalidation callback
+ * @nb: pointer to the gmap notifier block
+ */
+void gmap_unregister_ipte_notifier(struct gmap_notifier *nb)
+{
+	spin_lock(&gmap_notifier_lock);
+	list_del_init(&nb->list);
+	spin_unlock(&gmap_notifier_lock);
+}
+EXPORT_SYMBOL_GPL(gmap_unregister_ipte_notifier);
+
+/**
+ * gmap_ipte_notify - mark a range of ptes for invalidation notification
+ * @gmap: pointer to guest mapping meta data structure
+ * @address: virtual address in the guest address space
+ * @len: size of area
+ *
+ * Returns 0 if for each page in the given range a gmap mapping exists and
+ * the invalidation notification could be set. If the gmap mapping is missing
+ * for one or more pages -EFAULT is returned. If no memory could be allocated
+ * -ENOMEM is returned. This function establishes missing page table entries.
+ */
+int gmap_ipte_notify(struct gmap *gmap, unsigned long start, unsigned long len)
+{
+	unsigned long addr;
+	spinlock_t *ptl;
+	pte_t *ptep, entry;
+	pgste_t pgste;
+	int rc = 0;
+
+	if ((start & ~PAGE_MASK) || (len & ~PAGE_MASK))
+		return -EINVAL;
+	down_read(&gmap->mm->mmap_sem);
+	while (len) {
+		/* Convert gmap address and connect the page tables */
+		addr = __gmap_fault(start, gmap);
+		if (IS_ERR_VALUE(addr)) {
+			rc = addr;
+			break;
+		}
+		/* Get the page mapped */
+		if (get_user_pages(current, gmap->mm, addr, 1, 1, 0,
+				   NULL, NULL) != 1) {
+			rc = -EFAULT;
+			break;
+		}
+		/* Walk the process page table, lock and get pte pointer */
+		ptep = get_locked_pte(gmap->mm, addr, &ptl);
+		if (unlikely(!ptep))
+			continue;
+		/* Set notification bit in the pgste of the pte */
+		entry = *ptep;
+		if ((pte_val(entry) & (_PAGE_INVALID | _PAGE_RO)) == 0) {
+			pgste = pgste_get_lock(ptep);
+			pgste_val(pgste) |= RCP_IN_BIT;
+			pgste_set_unlock(ptep, pgste);
+			start += PAGE_SIZE;
+			len -= PAGE_SIZE;
+		}
+		spin_unlock(ptl);
+	}
+	up_read(&gmap->mm->mmap_sem);
+	return rc;
+}
+EXPORT_SYMBOL_GPL(gmap_ipte_notify);
+
+/**
+ * gmap_do_ipte_notify - call all invalidation callbacks for a specific pte.
+ * @mm: pointer to the process mm_struct
+ * @addr: virtual address in the process address space
+ * @pte: pointer to the page table entry
+ *
+ * This function is assumed to be called with the page table lock held
+ * for the pte to notify.
+ */
+void gmap_do_ipte_notify(struct mm_struct *mm, unsigned long addr, pte_t *pte)
+{
+	unsigned long segment_offset;
+	struct gmap_notifier *nb;
+	struct gmap_pgtable *mp;
+	struct gmap_rmap *rmap;
+	struct page *page;
+
+	segment_offset = ((unsigned long) pte) & (255 * sizeof(pte_t));
+	segment_offset = segment_offset * (4096 / sizeof(pte_t));
+	page = pfn_to_page(__pa(pte) >> PAGE_SHIFT);
+	mp = (struct gmap_pgtable *) page->index;
+	spin_lock(&gmap_notifier_lock);
+	list_for_each_entry(rmap, &mp->mapper, list) {
+		list_for_each_entry(nb, &gmap_notifier_list, list)
+			nb->notifier_call(rmap->gmap,
+					  rmap->vmaddr + segment_offset);
+	}
+	spin_unlock(&gmap_notifier_lock);
+}
+
 static inline unsigned long *page_table_alloc_pgste(struct mm_struct *mm,
 						    unsigned long vmaddr)
 {