Use mount_setattr() to set mount flags

Unlike the old mount(2) API, this lets us apply the same mount flags
recursively to a whole tree of mount points in a single operation,
avoiding time-of-check/time-of-use race conditions if the mount table
changes (#650) and improving performance when there are many mount
points (#384).

Fixes: #754

Signed-off-by: Guan Yunpeng <gitpull@protonmail.com>

Author: xxyzz

bind-mount.c

@@ -414,6 +414,36 @@ bind_mount (int           proc_fd,
       return BIND_MOUNT_ERROR_REOPEN_DEST;
     }
 
+  struct mount_attr attr = {
+    .attr_clr = 0,
+    .attr_set = MOUNT_ATTR_NOSUID,
+  };
+
+  if (!devices)
+    attr.attr_set |= MOUNT_ATTR_NODEV;
+
+  if (readonly)
+    attr.attr_set |= MOUNT_ATTR_RDONLY;
+
+  unsigned int setattr_flags = AT_EMPTY_PATH;
+
+  if (recursive)
+      setattr_flags |= AT_RECURSIVE;
+
+  if (mount_setattr_wrapper (dest_fd, "", setattr_flags, &attr, sizeof(attr)) == 0)
+    {
+      return BIND_MOUNT_SUCCESS;
+    }
+  else if (errno != ENOSYS)
+    {
+      if (failing_path != NULL)
+        *failing_path = steal_pointer (&resolved_dest);
+
+      return BIND_MOUNT_ERROR_MOUNT_SETATTR;
+    }
+
+  /* mount_setattr(2) isn't available, so we'll have to do this the hard way: */
+
   /* If we are in a case-insensitive filesystem, mountinfo might contain a
    * different case combination of the path we requested to mount.
    * This is due to the fact that the kernel, as of the beginning of 2021,
@@ -543,6 +573,10 @@ bind_mount_result_to_string (bind_mount_result res,
                             failing_path);
         break;
 
+      case BIND_MOUNT_ERROR_MOUNT_SETATTR:
+        string = xasprintf ("mount_setattr() failed at \"%s\"", failing_path);
+        break;
+
       case BIND_MOUNT_SUCCESS:
         string = xstrdup ("Success");
         break;
@@ -596,6 +630,7 @@ die_with_bind_result (bind_mount_result res,
           case BIND_MOUNT_ERROR_REOPEN_DEST:
           case BIND_MOUNT_ERROR_READLINK_DEST_PROC_FD:
           case BIND_MOUNT_ERROR_FIND_DEST_MOUNT:
+          case BIND_MOUNT_ERROR_MOUNT_SETATTR:
           case BIND_MOUNT_SUCCESS:
           default:
             fprintf (stderr, ": %s", strerror (saved_errno));

bind-mount.h

@@ -38,6 +38,7 @@ typedef enum
   BIND_MOUNT_ERROR_FIND_DEST_MOUNT,
   BIND_MOUNT_ERROR_REMOUNT_DEST,
   BIND_MOUNT_ERROR_REMOUNT_SUBMOUNT,
+  BIND_MOUNT_ERROR_MOUNT_SETATTR,
 } bind_mount_result;
 
 bind_mount_result bind_mount (int           proc_fd,

utils.c

@@ -1083,3 +1083,15 @@ strappend_escape_for_mount_options (StringBuilder *dest, const char *src)
       src++;
     }
 }
+
+int
+mount_setattr_wrapper (int dirfd, const char *path, unsigned int flags,
+                       struct mount_attr *attr, size_t size)
+{
+#ifdef __NR_mount_setattr
+  return syscall (__NR_mount_setattr, dirfd, path, flags, attr, size);
+#else
+  errno = ENOSYS;
+  return -1;
+#endif
+}

utils.h

@@ -215,3 +215,32 @@ void strappendf (StringBuilder *dest,
                  ...);
 void strappend_escape_for_mount_options (StringBuilder *dest,
                                          const char    *src);
+
+#ifndef MOUNT_ATTR_RDONLY
+struct mount_attr
+{
+  __u64 attr_set;
+  __u64 attr_clr;
+  __u64 propagation;
+  __u64 userns_fd;
+};
+
+#define MOUNT_ATTR_RDONLY       0x00000001
+#define MOUNT_ATTR_NOSUID       0x00000002
+#define MOUNT_ATTR_NODEV        0x00000004
+#define MOUNT_ATTR_NOEXEC       0x00000008
+#define MOUNT_ATTR__ATIME       0x00000070
+#define MOUNT_ATTR_RELATIME     0x00000000
+#define MOUNT_ATTR_NOATIME      0x00000010
+#define MOUNT_ATTR_STRICTATIME  0x00000020
+#define MOUNT_ATTR_NODIRATIME   0x00000080
+#define MOUNT_ATTR_IDMAP        0x00100000
+#define MOUNT_ATTR_NOSYMFOLLOW  0x00200000
+#endif
+
+#ifndef AT_RECURSIVE
+#define AT_RECURSIVE  0x8000
+#endif
+
+int mount_setattr_wrapper (int dirfd, const char *path, unsigned int flags,
+                           struct mount_attr *attr, size_t size);