xen/gntalloc,gntdev: Add unmap notify ioctl

This ioctl allows the users of a shared page to be notified when
the other end exits abnormally.

[v2: updated description in structs]
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>

drivers/xen/gntalloc.c

@@ -60,11 +60,13 @@
 #include <linux/uaccess.h>
 #include <linux/types.h>
 #include <linux/list.h>
+#include <linux/highmem.h>
 
 #include <xen/xen.h>
 #include <xen/page.h>
 #include <xen/grant_table.h>
 #include <xen/gntalloc.h>
+#include <xen/events.h>
 
 static int limit = 1024;
 module_param(limit, int, 0644);
@@ -75,6 +77,12 @@ static LIST_HEAD(gref_list);
 static DEFINE_SPINLOCK(gref_lock);
 static int gref_size;
 
+struct notify_info {
+	uint16_t pgoff:12;    /* Bits 0-11: Offset of the byte to clear */
+	uint16_t flags:2;     /* Bits 12-13: Unmap notification flags */
+	int event;            /* Port (event channel) to notify */
+};
+
 /* Metadata on a grant reference. */
 struct gntalloc_gref {
 	struct list_head next_gref;  /* list entry gref_list */
@@ -83,6 +91,7 @@ struct gntalloc_gref {
 	uint64_t file_index;         /* File offset for mmap() */
 	unsigned int users;          /* Use count - when zero, waiting on Xen */
 	grant_ref_t gref_id;         /* The grant reference number */
+	struct notify_info notify;   /* Unmap notification */
 };
 
 struct gntalloc_file_private_data {
@@ -164,6 +173,16 @@ undo:
 
 static void __del_gref(struct gntalloc_gref *gref)
 {
+	if (gref->notify.flags & UNMAP_NOTIFY_CLEAR_BYTE) {
+		uint8_t *tmp = kmap(gref->page);
+		tmp[gref->notify.pgoff] = 0;
+		kunmap(gref->page);
+	}
+	if (gref->notify.flags & UNMAP_NOTIFY_SEND_EVENT)
+		notify_remote_via_evtchn(gref->notify.event);
+
+	gref->notify.flags = 0;
+
 	if (gref->gref_id > 0) {
 		if (gnttab_query_foreign_access(gref->gref_id))
 			return;
@@ -349,6 +368,43 @@ dealloc_grant_out:
 	return rc;
 }
 
+static long gntalloc_ioctl_unmap_notify(struct gntalloc_file_private_data *priv,
+		void __user *arg)
+{
+	struct ioctl_gntalloc_unmap_notify op;
+	struct gntalloc_gref *gref;
+	uint64_t index;
+	int pgoff;
+	int rc;
+
+	if (copy_from_user(&op, arg, sizeof(op)))
+		return -EFAULT;
+
+	index = op.index & ~(PAGE_SIZE - 1);
+	pgoff = op.index & (PAGE_SIZE - 1);
+
+	spin_lock(&gref_lock);
+
+	gref = find_grefs(priv, index, 1);
+	if (!gref) {
+		rc = -ENOENT;
+		goto unlock_out;
+	}
+
+	if (op.action & ~(UNMAP_NOTIFY_CLEAR_BYTE|UNMAP_NOTIFY_SEND_EVENT)) {
+		rc = -EINVAL;
+		goto unlock_out;
+	}
+
+	gref->notify.flags = op.action;
+	gref->notify.pgoff = pgoff;
+	gref->notify.event = op.event_channel_port;
+	rc = 0;
+ unlock_out:
+	spin_unlock(&gref_lock);
+	return rc;
+}
+
 static long gntalloc_ioctl(struct file *filp, unsigned int cmd,
 		unsigned long arg)
 {
@@ -361,6 +417,9 @@ static long gntalloc_ioctl(struct file *filp, unsigned int cmd,
 	case IOCTL_GNTALLOC_DEALLOC_GREF:
 		return gntalloc_ioctl_dealloc(priv, (void __user *)arg);
 
+	case IOCTL_GNTALLOC_SET_UNMAP_NOTIFY:
+		return gntalloc_ioctl_unmap_notify(priv, (void __user *)arg);
+
 	default:
 		return -ENOIOCTLCMD;
 	}

drivers/xen/gntdev.c

@@ -37,6 +37,7 @@
 #include <xen/xen.h>
 #include <xen/grant_table.h>
 #include <xen/gntdev.h>
+#include <xen/events.h>
 #include <asm/xen/hypervisor.h>
 #include <asm/xen/hypercall.h>
 #include <asm/xen/page.h>
@@ -63,6 +64,13 @@ struct gntdev_priv {
 	struct mmu_notifier mn;
 };
 
+struct unmap_notify {
+	int flags;
+	/* Address relative to the start of the grant_map */
+	int addr;
+	int event;
+};
+
 struct grant_map {
 	struct list_head next;
 	struct vm_area_struct *vma;
@@ -71,6 +79,7 @@ struct grant_map {
 	int flags;
 	int is_mapped;
 	atomic_t users;
+	struct unmap_notify notify;
 	struct ioctl_gntdev_grant_ref *grants;
 	struct gnttab_map_grant_ref   *map_ops;
 	struct gnttab_unmap_grant_ref *unmap_ops;
@@ -165,7 +174,7 @@ static struct grant_map *gntdev_find_map_index(struct gntdev_priv *priv,
 	list_for_each_entry(map, &priv->maps, next) {
 		if (map->index != index)
 			continue;
-		if (map->count != count)
+		if (count && map->count != count)
 			continue;
 		return map;
 	}
@@ -184,6 +193,10 @@ static void gntdev_put_map(struct grant_map *map)
 
 	atomic_sub(map->count, &pages_mapped);
 
+	if (map->notify.flags & UNMAP_NOTIFY_SEND_EVENT) {
+		notify_remote_via_evtchn(map->notify.event);
+	}
+
 	if (map->pages) {
 		if (!use_ptemod)
 			unmap_grant_pages(map, 0, map->count);
@@ -274,6 +287,16 @@ static int unmap_grant_pages(struct grant_map *map, int offset, int pages)
 {
 	int i, err = 0;
 
+	if (map->notify.flags & UNMAP_NOTIFY_CLEAR_BYTE) {
+		int pgno = (map->notify.addr >> PAGE_SHIFT);
+		if (pgno >= offset && pgno < offset + pages) {
+			uint8_t *tmp = kmap(map->pages[pgno]);
+			tmp[map->notify.addr & (PAGE_SIZE-1)] = 0;
+			kunmap(map->pages[pgno]);
+			map->notify.flags &= ~UNMAP_NOTIFY_CLEAR_BYTE;
+		}
+	}
+
 	pr_debug("map %d+%d [%d+%d]\n", map->index, map->count, offset, pages);
 	err = gnttab_unmap_refs(map->unmap_ops + offset, map->pages, pages);
 	if (err)
@@ -519,6 +542,39 @@ static long gntdev_ioctl_get_offset_for_vaddr(struct gntdev_priv *priv,
 	return 0;
 }
 
+static long gntdev_ioctl_notify(struct gntdev_priv *priv, void __user *u)
+{
+	struct ioctl_gntdev_unmap_notify op;
+	struct grant_map *map;
+	int rc;
+
+	if (copy_from_user(&op, u, sizeof(op)))
+		return -EFAULT;
+
+	if (op.action & ~(UNMAP_NOTIFY_CLEAR_BYTE|UNMAP_NOTIFY_SEND_EVENT))
+		return -EINVAL;
+
+	spin_lock(&priv->lock);
+
+	list_for_each_entry(map, &priv->maps, next) {
+		uint64_t begin = map->index << PAGE_SHIFT;
+		uint64_t end = (map->index + map->count) << PAGE_SHIFT;
+		if (op.index >= begin && op.index < end)
+			goto found;
+	}
+	rc = -ENOENT;
+	goto unlock_out;
+
+ found:
+	map->notify.flags = op.action;
+	map->notify.addr = op.index - (map->index << PAGE_SHIFT);
+	map->notify.event = op.event_channel_port;
+	rc = 0;
+ unlock_out:
+	spin_unlock(&priv->lock);
+	return rc;
+}
+
 static long gntdev_ioctl(struct file *flip,
 			 unsigned int cmd, unsigned long arg)
 {
@@ -535,6 +591,9 @@ static long gntdev_ioctl(struct file *flip,
 	case IOCTL_GNTDEV_GET_OFFSET_FOR_VADDR:
 		return gntdev_ioctl_get_offset_for_vaddr(priv, ptr);
 
+	case IOCTL_GNTDEV_SET_UNMAP_NOTIFY:
+		return gntdev_ioctl_notify(priv, ptr);
+
 	default:
 		pr_debug("priv %p, unknown cmd %x\n", priv, cmd);
 		return -ENOIOCTLCMD;

include/xen/gntalloc.h

@@ -47,4 +47,36 @@ struct ioctl_gntalloc_dealloc_gref {
 	/* Number of references to unmap */
 	uint32_t count;
 };
+
+/*
+ * Sets up an unmap notification within the page, so that the other side can do
+ * cleanup if this side crashes. Required to implement cross-domain robust
+ * mutexes or close notification on communication channels.
+ *
+ * Each mapped page only supports one notification; multiple calls referring to
+ * the same page overwrite the previous notification. You must clear the
+ * notification prior to the IOCTL_GNTALLOC_DEALLOC_GREF if you do not want it
+ * to occur.
+ */
+#define IOCTL_GNTALLOC_SET_UNMAP_NOTIFY \
+_IOC(_IOC_NONE, 'G', 7, sizeof(struct ioctl_gntalloc_unmap_notify))
+struct ioctl_gntalloc_unmap_notify {
+	/* IN parameters */
+	/* Offset in the file descriptor for a byte within the page (same as
+	 * used in mmap). If using UNMAP_NOTIFY_CLEAR_BYTE, this is the byte to
+	 * be cleared. Otherwise, it can be any byte in the page whose
+	 * notification we are adjusting.
+	 */
+	uint64_t index;
+	/* Action(s) to take on unmap */
+	uint32_t action;
+	/* Event channel to notify */
+	uint32_t event_channel_port;
+};
+
+/* Clear (set to zero) the byte specified by index */
+#define UNMAP_NOTIFY_CLEAR_BYTE 0x1
+/* Send an interrupt on the indicated event channel */
+#define UNMAP_NOTIFY_SEND_EVENT 0x2
+
 #endif /* __LINUX_PUBLIC_GNTALLOC_H__ */

include/xen/gntdev.h

@@ -116,4 +116,35 @@ struct ioctl_gntdev_set_max_grants {
 	uint32_t count;
 };
 
+/*
+ * Sets up an unmap notification within the page, so that the other side can do
+ * cleanup if this side crashes. Required to implement cross-domain robust
+ * mutexes or close notification on communication channels.
+ *
+ * Each mapped page only supports one notification; multiple calls referring to
+ * the same page overwrite the previous notification. You must clear the
+ * notification prior to the IOCTL_GNTALLOC_DEALLOC_GREF if you do not want it
+ * to occur.
+ */
+#define IOCTL_GNTDEV_SET_UNMAP_NOTIFY \
+_IOC(_IOC_NONE, 'G', 7, sizeof(struct ioctl_gntdev_unmap_notify))
+struct ioctl_gntdev_unmap_notify {
+	/* IN parameters */
+	/* Offset in the file descriptor for a byte within the page (same as
+	 * used in mmap). If using UNMAP_NOTIFY_CLEAR_BYTE, this is the byte to
+	 * be cleared. Otherwise, it can be any byte in the page whose
+	 * notification we are adjusting.
+	 */
+	uint64_t index;
+	/* Action(s) to take on unmap */
+	uint32_t action;
+	/* Event channel to notify */
+	uint32_t event_channel_port;
+};
+
+/* Clear (set to zero) the byte specified by index */
+#define UNMAP_NOTIFY_CLEAR_BYTE 0x1
+/* Send an interrupt on the indicated event channel */
+#define UNMAP_NOTIFY_SEND_EVENT 0x2
+
 #endif /* __LINUX_PUBLIC_GNTDEV_H__ */