BACKPORT: copy_process(): don't use ksys_close() on cleanups

anon_inode_getfd() should be used *ONLY* in situations when we are
guaranteed to be past the last failure point (including copying the
descriptor number to userland, at that).  And ksys_close() should
not be used for cleanups at all.

anon_inode_getfile() is there for all nontrivial cases like that.
Just use that...

Bug: 254441685
Fixes: b3e5838 ("clone: add CLONE_PIDFD")
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Reviewed-by: Jann Horn <jannh@google.com>
Signed-off-by: Christian Brauner <christian@brauner.io>
(cherry picked from commit 6fd2fe4)
Signed-off-by: Lee Jones <joneslee@google.com>
Change-Id: Ib24efa6d9776ad1a6f9b8ca08e3aa8deefacfc20

kernel/fork.c

@@ -1712,31 +1712,6 @@ const struct file_operations pidfd_fops = {
 #endif
 };
 
-/**
- * pidfd_create() - Create a new pid file descriptor.
- *
- * @pid:  struct pid that the pidfd will reference
- *
- * This creates a new pid file descriptor with the O_CLOEXEC flag set.
- *
- * Note, that this function can only be called after the fd table has
- * been unshared to avoid leaking the pidfd to the new process.
- *
- * Return: On success, a cloexec pidfd is returned.
- *         On error, a negative errno number will be returned.
- */
-static int pidfd_create(struct pid *pid)
-{
-	int fd;
-
-	fd = anon_inode_getfd("[pidfd]", &pidfd_fops, get_pid(pid),
-			      O_RDWR | O_CLOEXEC);
-	if (fd < 0)
-		put_pid(pid);
-
-	return fd;
-}
-
 static void copy_oom_score_adj(u64 clone_flags, struct task_struct *tsk)
 {
 	/* Skip if kernel thread */
@@ -1778,6 +1753,7 @@ static __latent_entropy struct task_struct *copy_process(
 	int pidfd = -1, retval;
 	struct task_struct *p;
 	struct multiprocess_signals delayed;
+	struct file *pidfile = NULL;
 
 	/*
 	 * Don't allow sharing the root directory with processes in a different
@@ -2050,11 +2026,20 @@ static __latent_entropy struct task_struct *copy_process(
 	 * if the fd table isn't shared).
 	 */
 	if (clone_flags & CLONE_PIDFD) {
-		retval = pidfd_create(pid);
+		retval = get_unused_fd_flags(O_RDWR | O_CLOEXEC);
 		if (retval < 0)
 			goto bad_fork_free_pid;
 
 		pidfd = retval;
+
+		pidfile = anon_inode_getfile("[pidfd]", &pidfd_fops, pid,
+					      O_RDWR | O_CLOEXEC);
+		if (IS_ERR(pidfile)) {
+			put_unused_fd(pidfd);
+			goto bad_fork_free_pid;
+		}
+		get_pid(pid);	/* held by pidfile now */
+
 		retval = put_user(pidfd, parent_tidptr);
 		if (retval)
 			goto bad_fork_put_pidfd;
@@ -2166,6 +2151,9 @@ static __latent_entropy struct task_struct *copy_process(
 		goto bad_fork_cancel_cgroup;
 	}
 
+	/* past the last point of failure */
+	if (pidfile)
+		fd_install(pidfd, pidfile);
 
 	init_task_pid_links(p);
 	if (likely(p->pid)) {
@@ -2234,8 +2222,10 @@ static __latent_entropy struct task_struct *copy_process(
 bad_fork_cgroup_threadgroup_change_end:
 	cgroup_threadgroup_change_end(current);
 bad_fork_put_pidfd:
-	if (clone_flags & CLONE_PIDFD)
-		ksys_close(pidfd);
+	if (clone_flags & CLONE_PIDFD) {
+		fput(pidfile);
+		put_unused_fd(pidfd);
+	}
 bad_fork_free_pid:
 	if (pid != &init_struct_pid)
 		free_pid(pid);