fs: fix overflow in sys_mount() for in-kernel calls

sys_mount() reads/copies a whole page for its "type" parameter. When
do_mount_root() passes a kernel address that points to an object which is
smaller than a whole page, copy_mount_options() will happily go past this
memory object, possibly dereferencing "wild" pointers that could be in any
state (hence the kmemcheck warning, which shows that parts of the next
page are not even allocated).

(The likelihood of something going wrong here is pretty low -- first of
all this only applies to kernel calls to sys_mount(), which are mostly
found in the boot code. Secondly, I guess if the page was not mapped,
exact_copy_from_user() _would_ in fact handle it correctly because of its
access_ok(), etc. checks.)

But it is much nicer to avoid the dubious reads altogether, by stopping as
soon as we find a NUL byte. Is there a good reason why we can't do
something like this, using the already existing strndup_from_user()?

[[e-mail]: make copy_mount_string() static]
[AV: fix compat mount breakage, which involves undoing akpm's change above]

Reported-by: Ingo Molnar <[e-mail]>
Signed-off-by: Vegard Nossum <[e-mail]>
Cc: Al Viro <[e-mail]>
Cc: Pekka Enberg <[e-mail]>
Signed-off-by: Andrew Morton <[e-mail]>
Signed-off-by: al <[e-mail]>

fs/compat.c

···
 				 char __user * type, unsigned long flags,
 				 void __user * data)
 {
-	unsigned long type_page;
+	char *kernel_type;
 	unsigned long data_page;
-	unsigned long dev_page;
+	char *kernel_dev;
 	char *dir_page;
 	int retval;
 
-	retval = copy_mount_options (type, &type_page);
+	retval = copy_mount_string(type, &kernel_type);
 	if (retval < 0)
 		goto out;
 
···
 	if (IS_ERR(dir_page))
 		goto out1;
 
-	retval = copy_mount_options (dev_name, &dev_page);
+	retval = copy_mount_string(dev_name, &kernel_dev);
 	if (retval < 0)
 		goto out2;
 
-	retval = copy_mount_options (data, &data_page);
+	retval = copy_mount_options(data, &data_page);
 	if (retval < 0)
 		goto out3;
 
 	retval = -EINVAL;
 
-	if (type_page && data_page) {
-		if (!strcmp((char *)type_page, SMBFS_NAME)) {
+	if (kernel_type && data_page) {
+		if (!strcmp(kernel_type, SMBFS_NAME)) {
 			do_smb_super_data_conv((void *)data_page);
-		} else if (!strcmp((char *)type_page, NCPFS_NAME)) {
+		} else if (!strcmp(kernel_type, NCPFS_NAME)) {
 			do_ncp_super_data_conv((void *)data_page);
-		} else if (!strcmp((char *)type_page, NFS4_NAME)) {
+		} else if (!strcmp(kernel_type, NFS4_NAME)) {
 			if (do_nfs4_super_data_conv((void *) data_page))
 				goto out4;
 		}
 	}
 
-	retval = do_mount((char*)dev_page, dir_page, (char*)type_page,
+	retval = do_mount(kernel_dev, dir_page, kernel_type,
 			flags, (void*)data_page);
 
  out4:
 	free_page(data_page);
  out3:
-	free_page(dev_page);
+	kfree(kernel_dev);
  out2:
 	putname(dir_page);
  out1:
-	free_page(type_page);
+	kfree(kernel_type);
  out:
 	return retval;
 }

fs/internal.h

···
  * namespace.c
  */
 extern int copy_mount_options(const void __user *, unsigned long *);
+extern int copy_mount_string(const void __user *, char **);
 
 extern void free_vfsmnt(struct vfsmount *);
 extern struct vfsmount *alloc_vfsmnt(const char *);

fs/namespace.c

···
 {
 	struct vfsmount *mnt;
 
-	if (!type || !memchr(type, 0, PAGE_SIZE))
+	if (!type)
 		return -EINVAL;
 
 	/* we need capabilities... */
···
 	return 0;
 }
 
+int copy_mount_string(const void __user *data, char **where)
+{
+	char *tmp;
+
+	if (!data) {
+		*where = NULL;
+		return 0;
+	}
+
+	tmp = strndup_user(data, PAGE_SIZE);
+	if (IS_ERR(tmp))
+		return PTR_ERR(tmp);
+
+	*where = tmp;
+	return 0;
+}
+
 /*
  * Flags is a 32-bit value that allows up to 31 non-fs dependent flags to
  * be given to the mount() call (ie: read-only, no-dev, no-suid etc).
···
 	/* Basic sanity checks */
 
 	if (!dir_name || !*dir_name || !memchr(dir_name, 0, PAGE_SIZE))
-		return -EINVAL;
-	if (dev_name && !memchr(dev_name, 0, PAGE_SIZE))
 		return -EINVAL;
 
 	if (data_page)
···
 SYSCALL_DEFINE5(mount, char __user *, dev_name, char __user *, dir_name,
 		char __user *, type, unsigned long, flags, void __user *, data)
 {
-	int retval;
+	int ret;
+	char *kernel_type;
+	char *kernel_dir;
+	char *kernel_dev;
 	unsigned long data_page;
-	unsigned long type_page;
-	unsigned long dev_page;
-	char *dir_page;
 
-	retval = copy_mount_options(type, &type_page);
-	if (retval < 0)
-		return retval;
+	ret = copy_mount_string(type, &kernel_type);
+	if (ret < 0)
+		goto out_type;
 
-	dir_page = getname(dir_name);
-	retval = PTR_ERR(dir_page);
-	if (IS_ERR(dir_page))
-		goto out1;
+	kernel_dir = getname(dir_name);
+	if (IS_ERR(kernel_dir)) {
+		ret = PTR_ERR(kernel_dir);
+		goto out_dir;
+	}
 
-	retval = copy_mount_options(dev_name, &dev_page);
-	if (retval < 0)
-		goto out2;
+	ret = copy_mount_string(dev_name, &kernel_dev);
+	if (ret < 0)
+		goto out_dev;
 
-	retval = copy_mount_options(data, &data_page);
-	if (retval < 0)
-		goto out3;
+	ret = copy_mount_options(data, &data_page);
+	if (ret < 0)
+		goto out_data;
 
-	retval = do_mount((char *)dev_page, dir_page, (char *)type_page,
-			  flags, (void *)data_page);
+	ret = do_mount(kernel_dev, kernel_dir, kernel_type, flags,
+		(void *) data_page);
+
 	free_page(data_page);
-
-out3:
-	free_page(dev_page);
-out2:
-	putname(dir_page);
-out1:
-	free_page(type_page);
-	return retval;
+out_data:
+	kfree(kernel_dev);
+out_dev:
+	putname(kernel_dir);
+out_dir:
+	kfree(kernel_type);
+out_type:
+	return ret;
 }
 
 /*