[PATCH] UNTESTED UNTESTED DRAFT DRAFT

[Christian Brauner]

---
 include/linux/file.h | 20 ++++++++++++++++++++
 include/linux/pid.h  |  3 ++-
 kernel/fork.c        | 35 +++++++++++++++++++----------------
 kernel/pid.c         | 14 +++++++-------
 4 files changed, 48 insertions(+), 24 deletions(-)

diff --git a/include/linux/file.h b/include/linux/file.h
index 39704eae83e2..fdadaaa9f70b 100644
--- a/include/linux/file.h
+++ b/include/linux/file.h
@@ -39,6 +39,11 @@ struct fd {
 #define FDPUT_FPUT       1
 #define FDPUT_POS_UNLOCK 2
 
+struct fd_file {
+	struct file *file;
+	int fd;
+};
+
 static inline void fdput(struct fd fd)
 {
 	if (fd.flags & FDPUT_FPUT)
@@ -90,6 +95,21 @@ extern void put_unused_fd(unsigned int fd);
 
 extern void fd_install(unsigned int fd, struct file *file);
 
+static inline void fd_publish(struct fd_file *fdf)
+{
+	if (fdf->file)
+		fd_install(fdf->fd, fdf->file);
+}
+
+static inline void fd_discard(struct fd_file *fdf)
+{
+	if (fdf->file) {
+		fput(fdf->file);
+		put_unused_fd(fdf->fd);
+
+	}
+}
+
 extern int __receive_fd(struct file *file, int __user *ufd,
 			unsigned int o_flags);
 
diff --git a/include/linux/pid.h b/include/linux/pid.h
index b75de288a8c2..0b0695459508 100644
--- a/include/linux/pid.h
+++ b/include/linux/pid.h
@@ -75,12 +75,13 @@ extern struct pid init_struct_pid;
 extern const struct file_operations pidfd_fops;
 
 struct file;
+struct fd_file;
 
 extern struct pid *pidfd_pid(const struct file *file);
 struct pid *pidfd_get_pid(unsigned int fd, unsigned int *flags);
 struct task_struct *pidfd_get_task(int pidfd, unsigned int *flags);
 int pidfd_create(struct pid *pid, unsigned int flags);
-int pidfd_prepare(struct pid *pid, unsigned int flags, struct file **ret);
+int pidfd_prepare(struct pid *pid, unsigned int flags, struct fd_file *fdf);
 
 static inline struct pid *get_pid(struct pid *pid)
 {
diff --git a/kernel/fork.c b/kernel/fork.c
index bfe73db1c26c..2072d8cc91d2 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -1989,7 +1989,8 @@ const struct file_operations pidfd_fops = {
  *         error, a negative error code is returned from the function and the
  *         last argument remains unchanged.
  */
-static int __pidfd_prepare(struct pid *pid, unsigned int flags, struct file **ret)
+static int __pidfd_prepare(struct pid *pid, unsigned int flags,
+			   struct fd_file *fdf)
 {
 	int pidfd;
 	struct file *pidfd_file;
@@ -2008,8 +2009,11 @@ static int __pidfd_prepare(struct pid *pid, unsigned int flags, struct file **re
 		return PTR_ERR(pidfd_file);
 	}
 	get_pid(pid); /* held by pidfd_file now */
-	*ret = pidfd_file;
-	return pidfd;
+
+	fdf->file = pidfd_file;
+	fdf->fd = pidfd;
+
+	return 0;
 }
 
 /**
@@ -2038,12 +2042,12 @@ static int __pidfd_prepare(struct pid *pid, unsigned int flags, struct file **re
  *         error, a negative error code is returned from the function and the
  *         last argument remains unchanged.
  */
-int pidfd_prepare(struct pid *pid, unsigned int flags, struct file **ret)
+int pidfd_prepare(struct pid *pid, unsigned int flags, struct fd_file *fdf)
 {
 	if (!pid || !pid_has_task(pid, PIDTYPE_TGID))
 		return -EINVAL;
 
-	return __pidfd_prepare(pid, flags, ret);
+	return __pidfd_prepare(pid, flags, fdf);
 }
 
 static void __delayed_free_task(struct rcu_head *rhp)
@@ -2106,10 +2110,12 @@ __latent_entropy struct task_struct *copy_process(
 					int node,
 					struct kernel_clone_args *args)
 {
-	int pidfd = -1, retval;
+	int retval;
+	struct fd_file pidfd = {
+		.fd = -1,
+	};
 	struct task_struct *p;
 	struct multiprocess_signals delayed;
-	struct file *pidfile = NULL;
 	const u64 clone_flags = args->flags;
 	struct nsproxy *nsp = current->nsproxy;
 
@@ -2395,12 +2401,11 @@ __latent_entropy struct task_struct *copy_process(
 	 */
 	if (clone_flags & CLONE_PIDFD) {
 		/* Note that no task has been attached to @pid yet. */
-		retval = __pidfd_prepare(pid, O_RDWR | O_CLOEXEC, &pidfile);
+		retval = __pidfd_prepare(pid, O_RDWR | O_CLOEXEC, &pidfd);
 		if (retval < 0)
 			goto bad_fork_free_pid;
-		pidfd = retval;
 
-		retval = put_user(pidfd, args->pidfd);
+		retval = put_user(pidfd.fd, args->pidfd);
 		if (retval)
 			goto bad_fork_put_pidfd;
 	}
@@ -2584,8 +2589,8 @@ __latent_entropy struct task_struct *copy_process(
 	syscall_tracepoint_update(p);
 	write_unlock_irq(&tasklist_lock);
 
-	if (pidfile)
-		fd_install(pidfd, pidfile);
+	if (clone_flags & CLONE_PIDFD)
+		fd_publish(&pidfd);
 
 	proc_fork_connector(p);
 	sched_post_fork(p);
@@ -2605,10 +2610,8 @@ __latent_entropy struct task_struct *copy_process(
 	write_unlock_irq(&tasklist_lock);
 	cgroup_cancel_fork(p, args);
 bad_fork_put_pidfd:
-	if (clone_flags & CLONE_PIDFD) {
-		fput(pidfile);
-		put_unused_fd(pidfd);
-	}
+	if (clone_flags & CLONE_PIDFD)
+		fd_discard(&pidfd);
 bad_fork_free_pid:
 	if (pid != &init_struct_pid)
 		free_pid(pid);
diff --git a/kernel/pid.c b/kernel/pid.c
index f93954a0384d..6cd46002aca3 100644
--- a/kernel/pid.c
+++ b/kernel/pid.c
@@ -594,15 +594,15 @@ struct task_struct *pidfd_get_task(int pidfd, unsigned int *flags)
  */
 int pidfd_create(struct pid *pid, unsigned int flags)
 {
-	int pidfd;
-	struct file *pidfd_file;
+	int ret;
+	struct fd_file pidfd;
 
-	pidfd = pidfd_prepare(pid, flags, &pidfd_file);
-	if (pidfd < 0)
-		return pidfd;
+	ret = pidfd_prepare(pid, flags, &pidfd);
+	if (ret < 0)
+		return ret;
 
-	fd_install(pidfd, pidfd_file);
-	return pidfd;
+	fd_publish(&pidfd);
+	return pidfd.fd;
 }
 
 /**
-- 
2.34.1