Merge tag 'vfs-6.9.pidfd' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs

Pull pdfd updates from Christian Brauner:

 - Until now pidfds could only be created for thread-group leaders but
   not for threads. There was no technical reason for this. We simply
   had no users that needed support for this. Now we do have users that
   need support for this.

   This introduces a new PIDFD_THREAD flag for pidfd_open(). If that
   flag is set pidfd_open() creates a pidfd that refers to a specific
   thread.

   In addition, we now allow clone() and clone3() to be called with
   CLONE_PIDFD | CLONE_THREAD which wasn't possible before.

   A pidfd that refers to an individual thread differs from a pidfd that
   refers to a thread-group leader:

    (1) Pidfds are pollable. A task may poll a pidfd and get notified
        when the task has exited.

        For thread-group leader pidfds the polling task is woken if the
        thread-group is empty. In other words, if the thread-group
        leader task exits when there are still threads alive in its
        thread-group the polling task will not be woken when the
        thread-group leader exits but rather when the last thread in the
        thread-group exits.

        For thread-specific pidfds the polling task is woken if the
        thread exits.

    (2) Passing a thread-group leader pidfd to pidfd_send_signal() will
        generate thread-group directed signals like kill(2) does.

        Passing a thread-specific pidfd to pidfd_send_signal() will
        generate thread-specific signals like tgkill(2) does.

        The default scope of the signal is thus determined by the type
        of the pidfd.

        Since use-cases exist where the default scope of the provided
        pidfd needs to be overriden the following flags are added to
        pidfd_send_signal():

         - PIDFD_SIGNAL_THREAD
           Send a thread-specific signal.

         - PIDFD_SIGNAL_THREAD_GROUP
           Send a thread-group directed signal.

         - PIDFD_SIGNAL_PROCESS_GROUP
           Send a process-group directed signal.

        The scope change will only work if the struct pid is actually
        used for this scope.

        For example, in order to send a thread-group directed signal the
        provided pidfd must be used as a thread-group leader and
        similarly for PIDFD_SIGNAL_PROCESS_GROUP the struct pid must be
        used as a process group leader.

 - Move pidfds from the anonymous inode infrastructure to a tiny pseudo
   filesystem. This will unblock further work that we weren't able to do
   simply because of the very justified limitations of anonymous inodes.
   Moving pidfds to a tiny pseudo filesystem allows for statx on pidfds
   to become useful for the first time. They can now be compared by
   inode number which are unique for the system lifetime.

   Instead of stashing struct pid in file->private_data we can now stash
   it in inode->i_private. This makes it possible to introduce concepts
   that operate on a process once all file descriptors have been closed.
   A concrete example is kill-on-last-close. Another side-effect is that
   file->private_data is now freed up for per-file options for pidfds.

   Now, each struct pid will refer to a different inode but the same
   struct pid will refer to the same inode if it's opened multiple
   times. In contrast to now where each struct pid refers to the same
   inode.

   The tiny pseudo filesystem is not visible anywhere in userspace
   exactly like e.g., pipefs and sockfs. There's no lookup, there's no
   complex inode operations, nothing. Dentries and inodes are always
   deleted when the last pidfd is closed.

   We allocate a new inode and dentry for each struct pid and we reuse
   that inode and dentry for all pidfds that refer to the same struct
   pid. The code is entirely optional and fairly small. If it's not
   selected we fallback to anonymous inodes. Heavily inspired by nsfs.

   The dentry and inode allocation mechanism is moved into generic
   infrastructure that is now shared between nsfs and pidfs. The
   path_from_stashed() helper must be provided with a stashing location,
   an inode number, a mount, and the private data that is supposed to be
   used and it will provide a path that can be passed to dentry_open().

   The helper will try retrieve an existing dentry from the provided
   stashing location. If a valid dentry is found it is reused. If not a
   new one is allocated and we try to stash it in the provided location.
   If this fails we retry until we either find an existing dentry or the
   newly allocated dentry could be stashed. Subsequent openers of the
   same namespace or task are then able to reuse it.

 - Currently it is only possible to get notified when a task has exited,
   i.e., become a zombie and userspace gets notified with EPOLLIN. We
   now also support waiting until the task has been reaped, notifying
   userspace with EPOLLHUP.

 - Ensure that ESRCH is reported for getfd if a task is exiting instead
   of the confusing EBADF.

 - Various smaller cleanups to pidfd functions.

* tag 'vfs-6.9.pidfd' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs: (23 commits)
  libfs: improve path_from_stashed()
  libfs: add stashed_dentry_prune()
  libfs: improve path_from_stashed() helper
  pidfs: convert to path_from_stashed() helper
  nsfs: convert to path_from_stashed() helper
  libfs: add path_from_stashed()
  pidfd: add pidfs
  pidfd: move struct pidfd_fops
  pidfd: allow to override signal scope in pidfd_send_signal()
  pidfd: change pidfd_send_signal() to respect PIDFD_THREAD
  signal: fill in si_code in prepare_kill_siginfo()
  selftests: add ESRCH tests for pidfd_getfd()
  pidfd: getfd should always report ESRCH if a task is exiting
  pidfd: clone: allow CLONE_THREAD | CLONE_PIDFD together
  pidfd: exit: kill the no longer used thread_group_exited()
  pidfd: change do_notify_pidfd() to use __wake_up(poll_to_key(EPOLLIN))
  pid: kill the obsolete PIDTYPE_PID code in transfer_pid()
  pidfd: kill the no longer needed do_notify_pidfd() in de_thread()
  pidfd_poll: report POLLHUP when pid_task() == NULL
  pidfd: implement PIDFD_THREAD flag for pidfd_open()
  ...

fs/Kconfig

@@ -173,6 +173,13 @@ source "fs/proc/Kconfig"
 source "fs/kernfs/Kconfig"
 source "fs/sysfs/Kconfig"
 
+config FS_PID
+	bool "Pseudo filesystem for process file descriptors"
+	depends on 64BIT
+	default y
+	help
+	  Pidfs implements advanced features for process file descriptors.
+
 config TMPFS
 	bool "Tmpfs virtual memory file system support (former shm fs)"
 	depends on SHMEM

fs/Makefile

@@ -15,7 +15,7 @@ obj-y :=	open.o read_write.o file_table.o super.o \
 		pnode.o splice.o sync.o utimes.o d_path.o \
 		stack.o fs_struct.o statfs.o fs_pin.o nsfs.o \
 		fs_types.o fs_context.o fs_parser.o fsopen.o init.o \
-		kernel_read_file.o mnt_idmapping.o remap_range.o
+		kernel_read_file.o mnt_idmapping.o remap_range.o pidfs.o
 
 obj-$(CONFIG_BUFFER_HEAD)	+= buffer.o mpage.o
 obj-$(CONFIG_PROC_FS)		+= proc_namespace.o

fs/exec.c

@@ -1158,7 +1158,6 @@ static int de_thread(struct task_struct *tsk)
 
 		BUG_ON(leader->exit_state != EXIT_ZOMBIE);
 		leader->exit_state = EXIT_DEAD;
-
 		/*
 		 * We are going to release_task()->ptrace_unlink() silently,
 		 * the tracer can sleep in do_wait(). EXIT_DEAD guarantees

fs/internal.h

@@ -310,3 +310,10 @@ ssize_t __kernel_write_iter(struct file *file, struct iov_iter *from, loff_t *po
 struct mnt_idmap *alloc_mnt_idmap(struct user_namespace *mnt_userns);
 struct mnt_idmap *mnt_idmap_get(struct mnt_idmap *idmap);
 void mnt_idmap_put(struct mnt_idmap *idmap);
+struct stashed_operations {
+	void (*put_data)(void *data);
+	void (*init_inode)(struct inode *inode, void *data);
+};
+int path_from_stashed(struct dentry **stashed, unsigned long ino,
+		      struct vfsmount *mnt, void *data, struct path *path);
+void stashed_dentry_prune(struct dentry *dentry);

fs/libfs.c

@@ -23,6 +23,7 @@
 #include <linux/fsnotify.h>
 #include <linux/unicode.h>
 #include <linux/fscrypt.h>
+#include <linux/pidfs.h>
 
 #include <linux/uaccess.h>
 
@@ -1985,3 +1986,144 @@ struct timespec64 simple_inode_init_ts(struct inode *inode)
 	return ts;
 }
 EXPORT_SYMBOL(simple_inode_init_ts);
+
+static inline struct dentry *get_stashed_dentry(struct dentry *stashed)
+{
+	struct dentry *dentry;
+
+	guard(rcu)();
+	dentry = READ_ONCE(stashed);
+	if (!dentry)
+		return NULL;
+	if (!lockref_get_not_dead(&dentry->d_lockref))
+		return NULL;
+	return dentry;
+}
+
+static struct dentry *prepare_anon_dentry(struct dentry **stashed,
+					  unsigned long ino,
+					  struct super_block *sb,
+					  void *data)
+{
+	struct dentry *dentry;
+	struct inode *inode;
+	const struct stashed_operations *sops = sb->s_fs_info;
+
+	dentry = d_alloc_anon(sb);
+	if (!dentry)
+		return ERR_PTR(-ENOMEM);
+
+	inode = new_inode_pseudo(sb);
+	if (!inode) {
+		dput(dentry);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	inode->i_ino = ino;
+	inode->i_flags |= S_IMMUTABLE;
+	inode->i_mode = S_IFREG;
+	simple_inode_init_ts(inode);
+	sops->init_inode(inode, data);
+
+	/* Notice when this is changed. */
+	WARN_ON_ONCE(!S_ISREG(inode->i_mode));
+	WARN_ON_ONCE(!IS_IMMUTABLE(inode));
+
+	/* Store address of location where dentry's supposed to be stashed. */
+	dentry->d_fsdata = stashed;
+
+	/* @data is now owned by the fs */
+	d_instantiate(dentry, inode);
+	return dentry;
+}
+
+static struct dentry *stash_dentry(struct dentry **stashed,
+				   struct dentry *dentry)
+{
+	guard(rcu)();
+	for (;;) {
+		struct dentry *old;
+
+		/* Assume any old dentry was cleared out. */
+		old = cmpxchg(stashed, NULL, dentry);
+		if (likely(!old))
+			return dentry;
+
+		/* Check if somebody else installed a reusable dentry. */
+		if (lockref_get_not_dead(&old->d_lockref))
+			return old;
+
+		/* There's an old dead dentry there, try to take it over. */
+		if (likely(try_cmpxchg(stashed, &old, dentry)))
+			return dentry;
+	}
+}
+
+/**
+ * path_from_stashed - create path from stashed or new dentry
+ * @stashed:    where to retrieve or stash dentry
+ * @ino:        inode number to use
+ * @mnt:        mnt of the filesystems to use
+ * @data:       data to store in inode->i_private
+ * @path:       path to create
+ *
+ * The function tries to retrieve a stashed dentry from @stashed. If the dentry
+ * is still valid then it will be reused. If the dentry isn't able the function
+ * will allocate a new dentry and inode. It will then check again whether it
+ * can reuse an existing dentry in case one has been added in the meantime or
+ * update @stashed with the newly added dentry.
+ *
+ * Special-purpose helper for nsfs and pidfs.
+ *
+ * Return: On success zero and on failure a negative error is returned.
+ */
+int path_from_stashed(struct dentry **stashed, unsigned long ino,
+		      struct vfsmount *mnt, void *data, struct path *path)
+{
+	struct dentry *dentry;
+	const struct stashed_operations *sops = mnt->mnt_sb->s_fs_info;
+
+	/* See if dentry can be reused. */
+	path->dentry = get_stashed_dentry(*stashed);
+	if (path->dentry) {
+		sops->put_data(data);
+		goto out_path;
+	}
+
+	/* Allocate a new dentry. */
+	dentry = prepare_anon_dentry(stashed, ino, mnt->mnt_sb, data);
+	if (IS_ERR(dentry)) {
+		sops->put_data(data);
+		return PTR_ERR(dentry);
+	}
+
+	/* Added a new dentry. @data is now owned by the filesystem. */
+	path->dentry = stash_dentry(stashed, dentry);
+	if (path->dentry != dentry)
+		dput(dentry);
+
+out_path:
+	WARN_ON_ONCE(path->dentry->d_fsdata != stashed);
+	WARN_ON_ONCE(d_inode(path->dentry)->i_private != data);
+	path->mnt = mntget(mnt);
+	return 0;
+}
+
+void stashed_dentry_prune(struct dentry *dentry)
+{
+	struct dentry **stashed = dentry->d_fsdata;
+	struct inode *inode = d_inode(dentry);
+
+	if (WARN_ON_ONCE(!stashed))
+		return;
+
+	if (!inode)
+		return;
+
+	/*
+	 * Only replace our own @dentry as someone else might've
+	 * already cleared out @dentry and stashed their own
+	 * dentry in there.
+	 */
+	cmpxchg(stashed, dentry, NULL);
+}

fs/nsfs.c

@@ -27,26 +27,17 @@ static const struct file_operations ns_file_operations = {
 static char *ns_dname(struct dentry *dentry, char *buffer, int buflen)
 {
 	struct inode *inode = d_inode(dentry);
-	const struct proc_ns_operations *ns_ops = dentry->d_fsdata;
+	struct ns_common *ns = inode->i_private;
+	const struct proc_ns_operations *ns_ops = ns->ops;
 
 	return dynamic_dname(buffer, buflen, "%s:[%lu]",
 		ns_ops->name, inode->i_ino);
 }
 
-static void ns_prune_dentry(struct dentry *dentry)
-{
-	struct inode *inode = d_inode(dentry);
-	if (inode) {
-		struct ns_common *ns = inode->i_private;
-		atomic_long_set(&ns->stashed, 0);
-	}
-}
-
-const struct dentry_operations ns_dentry_operations =
-{
-	.d_prune	= ns_prune_dentry,
+const struct dentry_operations ns_dentry_operations = {
 	.d_delete	= always_delete_dentry,
 	.d_dname	= ns_dname,
+	.d_prune	= stashed_dentry_prune,
 };
 
 static void nsfs_evict(struct inode *inode)
@@ -56,67 +47,16 @@ static void nsfs_evict(struct inode *inode)
 	ns->ops->put(ns);
 }
 
-static int __ns_get_path(struct path *path, struct ns_common *ns)
-{
-	struct vfsmount *mnt = nsfs_mnt;
-	struct dentry *dentry;
-	struct inode *inode;
-	unsigned long d;
-
-	rcu_read_lock();
-	d = atomic_long_read(&ns->stashed);
-	if (!d)
-		goto slow;
-	dentry = (struct dentry *)d;
-	if (!lockref_get_not_dead(&dentry->d_lockref))
-		goto slow;
-	rcu_read_unlock();
-	ns->ops->put(ns);
-got_it:
-	path->mnt = mntget(mnt);
-	path->dentry = dentry;
-	return 0;
-slow:
-	rcu_read_unlock();
-	inode = new_inode_pseudo(mnt->mnt_sb);
-	if (!inode) {
-		ns->ops->put(ns);
-		return -ENOMEM;
-	}
-	inode->i_ino = ns->inum;
-	simple_inode_init_ts(inode);
-	inode->i_flags |= S_IMMUTABLE;
-	inode->i_mode = S_IFREG | S_IRUGO;
-	inode->i_fop = &ns_file_operations;
-	inode->i_private = ns;
-
-	dentry = d_make_root(inode);	/* not the normal use, but... */
-	if (!dentry)
-		return -ENOMEM;
-	dentry->d_fsdata = (void *)ns->ops;
-	d = atomic_long_cmpxchg(&ns->stashed, 0, (unsigned long)dentry);
-	if (d) {
-		d_delete(dentry);	/* make sure ->d_prune() does nothing */
-		dput(dentry);
-		cpu_relax();
-		return -EAGAIN;
-	}
-	goto got_it;
-}
-
 int ns_get_path_cb(struct path *path, ns_get_path_helper_t *ns_get_cb,
 		     void *private_data)
 {
-	int ret;
+	struct ns_common *ns;
 
-	do {
-		struct ns_common *ns = ns_get_cb(private_data);
-		if (!ns)
-			return -ENOENT;
-		ret = __ns_get_path(path, ns);
-	} while (ret == -EAGAIN);
+	ns = ns_get_cb(private_data);
+	if (!ns)
+		return -ENOENT;
 
-	return ret;
+	return path_from_stashed(&ns->stashed, ns->inum, nsfs_mnt, ns, path);
 }
 
 struct ns_get_path_task_args {
@@ -146,6 +86,7 @@ int open_related_ns(struct ns_common *ns,
 		   struct ns_common *(*get_ns)(struct ns_common *ns))
 {
 	struct path path = {};
+	struct ns_common *relative;
 	struct file *f;
 	int err;
 	int fd;
@@ -154,19 +95,15 @@ int open_related_ns(struct ns_common *ns,
 	if (fd < 0)
 		return fd;
 
-	do {
-		struct ns_common *relative;
-
-		relative = get_ns(ns);
-		if (IS_ERR(relative)) {
-			put_unused_fd(fd);
-			return PTR_ERR(relative);
-		}
-
-		err = __ns_get_path(&path, relative);
-	} while (err == -EAGAIN);
+	relative = get_ns(ns);
+	if (IS_ERR(relative)) {
+		put_unused_fd(fd);
+		return PTR_ERR(relative);
+	}
 
-	if (err) {
+	err = path_from_stashed(&relative->stashed, relative->inum, nsfs_mnt,
+				relative, &path);
+	if (err < 0) {
 		put_unused_fd(fd);
 		return err;
 	}
@@ -249,7 +186,8 @@ bool ns_match(const struct ns_common *ns, dev_t dev, ino_t ino)
 static int nsfs_show_path(struct seq_file *seq, struct dentry *dentry)
 {
 	struct inode *inode = d_inode(dentry);
-	const struct proc_ns_operations *ns_ops = dentry->d_fsdata;
+	const struct ns_common *ns = inode->i_private;
+	const struct proc_ns_operations *ns_ops = ns->ops;
 
 	seq_printf(seq, "%s:[%lu]", ns_ops->name, inode->i_ino);
 	return 0;
@@ -261,6 +199,24 @@ static const struct super_operations nsfs_ops = {
 	.show_path = nsfs_show_path,
 };
 
+static void nsfs_init_inode(struct inode *inode, void *data)
+{
+	inode->i_private = data;
+	inode->i_mode |= S_IRUGO;
+	inode->i_fop = &ns_file_operations;
+}
+
+static void nsfs_put_data(void *data)
+{
+	struct ns_common *ns = data;
+	ns->ops->put(ns);
+}
+
+static const struct stashed_operations nsfs_stashed_ops = {
+	.init_inode = nsfs_init_inode,
+	.put_data = nsfs_put_data,
+};
+
 static int nsfs_init_fs_context(struct fs_context *fc)
 {
 	struct pseudo_fs_context *ctx = init_pseudo(fc, NSFS_MAGIC);
@@ -268,6 +224,7 @@ static int nsfs_init_fs_context(struct fs_context *fc)
 		return -ENOMEM;
 	ctx->ops = &nsfs_ops;
 	ctx->dops = &ns_dentry_operations;
+	fc->s_fs_info = (void *)&nsfs_stashed_ops;
 	return 0;
 }
 

fs/pidfs.c

+// SPDX-License-Identifier: GPL-2.0
+#include <linux/anon_inodes.h>
+#include <linux/file.h>
+#include <linux/fs.h>
+#include <linux/magic.h>
+#include <linux/mount.h>
+#include <linux/pid.h>
+#include <linux/pidfs.h>
+#include <linux/pid_namespace.h>
+#include <linux/poll.h>
+#include <linux/proc_fs.h>
+#include <linux/proc_ns.h>
+#include <linux/pseudo_fs.h>
+#include <linux/seq_file.h>
+#include <uapi/linux/pidfd.h>
+
+#include "internal.h"
+
+static int pidfd_release(struct inode *inode, struct file *file)
+{
+#ifndef CONFIG_FS_PID
+	struct pid *pid = file->private_data;
+
+	file->private_data = NULL;
+	put_pid(pid);
+#endif
+	return 0;
+}
+
+#ifdef CONFIG_PROC_FS
+/**
+ * pidfd_show_fdinfo - print information about a pidfd
+ * @m: proc fdinfo file
+ * @f: file referencing a pidfd
+ *
+ * Pid:
+ * This function will print the pid that a given pidfd refers to in the
+ * pid namespace of the procfs instance.
+ * If the pid namespace of the process is not a descendant of the pid
+ * namespace of the procfs instance 0 will be shown as its pid. This is
+ * similar to calling getppid() on a process whose parent is outside of
+ * its pid namespace.
+ *
+ * NSpid:
+ * If pid namespaces are supported then this function will also print
+ * the pid of a given pidfd refers to for all descendant pid namespaces
+ * starting from the current pid namespace of the instance, i.e. the
+ * Pid field and the first entry in the NSpid field will be identical.
+ * If the pid namespace of the process is not a descendant of the pid
+ * namespace of the procfs instance 0 will be shown as its first NSpid
+ * entry and no others will be shown.
+ * Note that this differs from the Pid and NSpid fields in
+ * /proc/<pid>/status where Pid and NSpid are always shown relative to
+ * the  pid namespace of the procfs instance. The difference becomes
+ * obvious when sending around a pidfd between pid namespaces from a
+ * different branch of the tree, i.e. where no ancestral relation is
+ * present between the pid namespaces:
+ * - create two new pid namespaces ns1 and ns2 in the initial pid
+ *   namespace (also take care to create new mount namespaces in the
+ *   new pid namespace and mount procfs)
+ * - create a process with a pidfd in ns1
+ * - send pidfd from ns1 to ns2
+ * - read /proc/self/fdinfo/<pidfd> and observe that both Pid and NSpid
+ *   have exactly one entry, which is 0
+ */
+static void pidfd_show_fdinfo(struct seq_file *m, struct file *f)
+{
+	struct pid *pid = pidfd_pid(f);
+	struct pid_namespace *ns;
+	pid_t nr = -1;
+
+	if (likely(pid_has_task(pid, PIDTYPE_PID))) {
+		ns = proc_pid_ns(file_inode(m->file)->i_sb);
+		nr = pid_nr_ns(pid, ns);
+	}
+
+	seq_put_decimal_ll(m, "Pid:\t", nr);
+
+#ifdef CONFIG_PID_NS
+	seq_put_decimal_ll(m, "\nNSpid:\t", nr);
+	if (nr > 0) {
+		int i;
+
+		/* If nr is non-zero it means that 'pid' is valid and that
+		 * ns, i.e. the pid namespace associated with the procfs
+		 * instance, is in the pid namespace hierarchy of pid.
+		 * Start at one below the already printed level.
+		 */
+		for (i = ns->level + 1; i <= pid->level; i++)
+			seq_put_decimal_ll(m, "\t", pid->numbers[i].nr);
+	}
+#endif
+	seq_putc(m, '\n');
+}
+#endif
+
+/*
+ * Poll support for process exit notification.
+ */
+static __poll_t pidfd_poll(struct file *file, struct poll_table_struct *pts)
+{
+	struct pid *pid = pidfd_pid(file);
+	bool thread = file->f_flags & PIDFD_THREAD;
+	struct task_struct *task;
+	__poll_t poll_flags = 0;
+
+	poll_wait(file, &pid->wait_pidfd, pts);
+	/*
+	 * Depending on PIDFD_THREAD, inform pollers when the thread
+	 * or the whole thread-group exits.
+	 */
+	guard(rcu)();
+	task = pid_task(pid, PIDTYPE_PID);
+	if (!task)
+		poll_flags = EPOLLIN | EPOLLRDNORM | EPOLLHUP;
+	else if (task->exit_state && (thread || thread_group_empty(task)))
+		poll_flags = EPOLLIN | EPOLLRDNORM;
+
+	return poll_flags;
+}
+
+static const struct file_operations pidfs_file_operations = {
+	.release	= pidfd_release,
+	.poll		= pidfd_poll,
+#ifdef CONFIG_PROC_FS
+	.show_fdinfo	= pidfd_show_fdinfo,
+#endif
+};
+
+struct pid *pidfd_pid(const struct file *file)
+{
+	if (file->f_op != &pidfs_file_operations)
+		return ERR_PTR(-EBADF);
+#ifdef CONFIG_FS_PID
+	return file_inode(file)->i_private;
+#else
+	return file->private_data;
+#endif
+}
+
+#ifdef CONFIG_FS_PID
+static struct vfsmount *pidfs_mnt __ro_after_init;
+
+/*
+ * The vfs falls back to simple_setattr() if i_op->setattr() isn't
+ * implemented. Let's reject it completely until we have a clean
+ * permission concept for pidfds.
+ */
+static int pidfs_setattr(struct mnt_idmap *idmap, struct dentry *dentry,
+			 struct iattr *attr)
+{
+	return -EOPNOTSUPP;
+}
+
+static int pidfs_getattr(struct mnt_idmap *idmap, const struct path *path,
+			 struct kstat *stat, u32 request_mask,
+			 unsigned int query_flags)
+{
+	struct inode *inode = d_inode(path->dentry);
+
+	generic_fillattr(&nop_mnt_idmap, request_mask, inode, stat);
+	return 0;
+}
+
+static const struct inode_operations pidfs_inode_operations = {
+	.getattr = pidfs_getattr,
+	.setattr = pidfs_setattr,
+};
+
+static void pidfs_evict_inode(struct inode *inode)
+{
+	struct pid *pid = inode->i_private;
+
+	clear_inode(inode);
+	put_pid(pid);
+}
+
+static const struct super_operations pidfs_sops = {
+	.drop_inode	= generic_delete_inode,
+	.evict_inode	= pidfs_evict_inode,
+	.statfs		= simple_statfs,
+};
+
+static char *pidfs_dname(struct dentry *dentry, char *buffer, int buflen)
+{
+	return dynamic_dname(buffer, buflen, "pidfd:[%lu]",
+			     d_inode(dentry)->i_ino);
+}
+
+static const struct dentry_operations pidfs_dentry_operations = {
+	.d_delete	= always_delete_dentry,
+	.d_dname	= pidfs_dname,
+	.d_prune	= stashed_dentry_prune,
+};
+
+static void pidfs_init_inode(struct inode *inode, void *data)
+{
+	inode->i_private = data;
+	inode->i_flags |= S_PRIVATE;
+	inode->i_mode |= S_IRWXU;
+	inode->i_op = &pidfs_inode_operations;
+	inode->i_fop = &pidfs_file_operations;
+}
+
+static void pidfs_put_data(void *data)
+{
+	struct pid *pid = data;
+	put_pid(pid);
+}
+
+static const struct stashed_operations pidfs_stashed_ops = {
+	.init_inode = pidfs_init_inode,
+	.put_data = pidfs_put_data,
+};
+
+static int pidfs_init_fs_context(struct fs_context *fc)
+{
+	struct pseudo_fs_context *ctx;
+
+	ctx = init_pseudo(fc, PID_FS_MAGIC);
+	if (!ctx)
+		return -ENOMEM;
+
+	ctx->ops = &pidfs_sops;
+	ctx->dops = &pidfs_dentry_operations;
+	fc->s_fs_info = (void *)&pidfs_stashed_ops;
+	return 0;
+}
+
+static struct file_system_type pidfs_type = {
+	.name			= "pidfs",
+	.init_fs_context	= pidfs_init_fs_context,
+	.kill_sb		= kill_anon_super,
+};
+
+struct file *pidfs_alloc_file(struct pid *pid, unsigned int flags)
+{
+
+	struct file *pidfd_file;
+	struct path path;
+	int ret;
+
+	/*
+	* Inode numbering for pidfs start at RESERVED_PIDS + 1.
+	* This avoids collisions with the root inode which is 1
+	* for pseudo filesystems.
+	 */
+	ret = path_from_stashed(&pid->stashed, pid->ino, pidfs_mnt,
+				get_pid(pid), &path);
+	if (ret < 0)
+		return ERR_PTR(ret);
+
+	pidfd_file = dentry_open(&path, flags, current_cred());
+	path_put(&path);
+	return pidfd_file;
+}
+
+void __init pidfs_init(void)
+{
+	pidfs_mnt = kern_mount(&pidfs_type);
+	if (IS_ERR(pidfs_mnt))
+		panic("Failed to mount pidfs pseudo filesystem");
+}
+
+bool is_pidfs_sb(const struct super_block *sb)
+{
+	return sb == pidfs_mnt->mnt_sb;
+}
+
+#else /* !CONFIG_FS_PID */
+
+struct file *pidfs_alloc_file(struct pid *pid, unsigned int flags)
+{
+	struct file *pidfd_file;
+
+	pidfd_file = anon_inode_getfile("[pidfd]", &pidfs_file_operations, pid,
+					flags | O_RDWR);
+	if (IS_ERR(pidfd_file))
+		return pidfd_file;
+
+	get_pid(pid);
+	return pidfd_file;
+}
+
+void __init pidfs_init(void) { }
+bool is_pidfs_sb(const struct super_block *sb)
+{
+	return false;
+}
+#endif

include/linux/ns_common.h

@@ -7,7 +7,7 @@
 struct proc_ns_operations;
 
 struct ns_common {
-	atomic_long_t stashed;
+	struct dentry *stashed;
 	const struct proc_ns_operations *ops;
 	unsigned int inum;
 	refcount_t count;

include/linux/pid.h

@@ -55,6 +55,10 @@ struct pid
 	refcount_t count;
 	unsigned int level;
 	spinlock_t lock;
+#ifdef CONFIG_FS_PID
+	struct dentry *stashed;
+	unsigned long ino;
+#endif
 	/* lists of tasks that use this pid */
 	struct hlist_head tasks[PIDTYPE_MAX];
 	struct hlist_head inodes;
@@ -66,15 +70,13 @@ struct pid
 
 extern struct pid init_struct_pid;
 
-extern const struct file_operations pidfd_fops;
-
 struct file;
 
-extern struct pid *pidfd_pid(const struct file *file);
+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);
+void do_notify_pidfd(struct task_struct *task);
 
 static inline struct pid *get_pid(struct pid *pid)
 {

include/linux/pidfs.h

+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _LINUX_PID_FS_H
+#define _LINUX_PID_FS_H
+
+struct file *pidfs_alloc_file(struct pid *pid, unsigned int flags);
+void __init pidfs_init(void);
+bool is_pidfs_sb(const struct super_block *sb);
+
+#endif /* _LINUX_PID_FS_H */

include/linux/proc_ns.h

@@ -66,7 +66,7 @@ static inline void proc_free_inum(unsigned int inum) {}
 
 static inline int ns_alloc_inum(struct ns_common *ns)
 {
-	atomic_long_set(&ns->stashed, 0);
+	WRITE_ONCE(ns->stashed, NULL);
 	return proc_alloc_inum(&ns->inum);
 }
 

include/linux/sched/signal.h

@@ -735,8 +735,6 @@ static inline int thread_group_empty(struct task_struct *p)
 #define delay_group_leader(p) \
 		(thread_group_leader(p) && !thread_group_empty(p))
 
-extern bool thread_group_exited(struct pid *pid);
-
 extern struct sighand_struct *__lock_task_sighand(struct task_struct *task,
 							unsigned long *flags);
 

include/uapi/linux/magic.h

@@ -101,5 +101,6 @@
 #define DMA_BUF_MAGIC		0x444d4142	/* "DMAB" */
 #define DEVMEM_MAGIC		0x454d444d	/* "DMEM" */
 #define SECRETMEM_MAGIC		0x5345434d	/* "SECM" */
+#define PID_FS_MAGIC		0x50494446	/* "PIDF" */
 
 #endif /* __LINUX_MAGIC_H__ */

include/uapi/linux/pidfd.h

@@ -7,6 +7,12 @@
 #include <linux/fcntl.h>
 
 /* Flags for pidfd_open().  */
-#define PIDFD_NONBLOCK O_NONBLOCK
+#define PIDFD_NONBLOCK	O_NONBLOCK
+#define PIDFD_THREAD	O_EXCL
+
+/* Flags for pidfd_send_signal(). */
+#define PIDFD_SIGNAL_THREAD		(1UL << 0)
+#define PIDFD_SIGNAL_THREAD_GROUP	(1UL << 1)
+#define PIDFD_SIGNAL_PROCESS_GROUP	(1UL << 2)
 
 #endif /* _UAPI_LINUX_PIDFD_H */

init/main.c

@@ -99,6 +99,7 @@
 #include <linux/init_syscalls.h>
 #include <linux/stackdepot.h>
 #include <linux/randomize_kstack.h>
+#include <linux/pidfs.h>
 #include <net/net_namespace.h>
 
 #include <asm/io.h>
@@ -1059,6 +1060,7 @@ void start_kernel(void)
 	seq_file_init();
 	proc_root_init();
 	nsfs_init();
+	pidfs_init();
 	cpuset_init();
 	cgroup_init();
 	taskstats_init_early();

kernel/exit.c

@@ -739,6 +739,13 @@ static void exit_notify(struct task_struct *tsk, int group_dead)
 		kill_orphaned_pgrp(tsk->group_leader, NULL);
 
 	tsk->exit_state = EXIT_ZOMBIE;
+	/*
+	 * sub-thread or delay_group_leader(), wake up the
+	 * PIDFD_THREAD waiters.
+	 */
+	if (!thread_group_empty(tsk))
+		do_notify_pidfd(tsk);
+
 	if (unlikely(tsk->ptrace)) {
 		int sig = thread_group_leader(tsk) &&
 				thread_group_empty(tsk) &&
@@ -1889,30 +1896,6 @@ COMPAT_SYSCALL_DEFINE5(waitid,
 }
 #endif
 
-/**
- * thread_group_exited - check that a thread group has exited
- * @pid: tgid of thread group to be checked.
- *
- * Test if the thread group represented by tgid has exited (all
- * threads are zombies, dead or completely gone).
- *
- * Return: true if the thread group has exited. false otherwise.
- */
-bool thread_group_exited(struct pid *pid)
-{
-	struct task_struct *task;
-	bool exited;
-
-	rcu_read_lock();
-	task = pid_task(pid, PIDTYPE_PID);
-	exited = !task ||
-		(READ_ONCE(task->exit_state) && thread_group_empty(task));
-	rcu_read_unlock();
-
-	return exited;
-}
-EXPORT_SYMBOL(thread_group_exited);
-
 /*
  * This needs to be __function_aligned as GCC implicitly makes any
  * implementation of abort() cold and drops alignment specified by

kernel/fork.c

@@ -101,6 +101,8 @@
 #include <linux/user_events.h>
 #include <linux/iommu.h>
 #include <linux/rseq.h>
+#include <uapi/linux/pidfd.h>
+#include <linux/pidfs.h>
 
 #include <asm/pgalloc.h>
 #include <linux/uaccess.h>
@@ -1985,119 +1987,6 @@ static inline void rcu_copy_process(struct task_struct *p)
 #endif /* #ifdef CONFIG_TASKS_TRACE_RCU */
 }
 
-struct pid *pidfd_pid(const struct file *file)
-{
-	if (file->f_op == &pidfd_fops)
-		return file->private_data;
-
-	return ERR_PTR(-EBADF);
-}
-
-static int pidfd_release(struct inode *inode, struct file *file)
-{
-	struct pid *pid = file->private_data;
-
-	file->private_data = NULL;
-	put_pid(pid);
-	return 0;
-}
-
-#ifdef CONFIG_PROC_FS
-/**
- * pidfd_show_fdinfo - print information about a pidfd
- * @m: proc fdinfo file
- * @f: file referencing a pidfd
- *
- * Pid:
- * This function will print the pid that a given pidfd refers to in the
- * pid namespace of the procfs instance.
- * If the pid namespace of the process is not a descendant of the pid
- * namespace of the procfs instance 0 will be shown as its pid. This is
- * similar to calling getppid() on a process whose parent is outside of
- * its pid namespace.
- *
- * NSpid:
- * If pid namespaces are supported then this function will also print
- * the pid of a given pidfd refers to for all descendant pid namespaces
- * starting from the current pid namespace of the instance, i.e. the
- * Pid field and the first entry in the NSpid field will be identical.
- * If the pid namespace of the process is not a descendant of the pid
- * namespace of the procfs instance 0 will be shown as its first NSpid
- * entry and no others will be shown.
- * Note that this differs from the Pid and NSpid fields in
- * /proc/<pid>/status where Pid and NSpid are always shown relative to
- * the  pid namespace of the procfs instance. The difference becomes
- * obvious when sending around a pidfd between pid namespaces from a
- * different branch of the tree, i.e. where no ancestral relation is
- * present between the pid namespaces:
- * - create two new pid namespaces ns1 and ns2 in the initial pid
- *   namespace (also take care to create new mount namespaces in the
- *   new pid namespace and mount procfs)
- * - create a process with a pidfd in ns1
- * - send pidfd from ns1 to ns2
- * - read /proc/self/fdinfo/<pidfd> and observe that both Pid and NSpid
- *   have exactly one entry, which is 0
- */
-static void pidfd_show_fdinfo(struct seq_file *m, struct file *f)
-{
-	struct pid *pid = f->private_data;
-	struct pid_namespace *ns;
-	pid_t nr = -1;
-
-	if (likely(pid_has_task(pid, PIDTYPE_PID))) {
-		ns = proc_pid_ns(file_inode(m->file)->i_sb);
-		nr = pid_nr_ns(pid, ns);
-	}
-
-	seq_put_decimal_ll(m, "Pid:\t", nr);
-
-#ifdef CONFIG_PID_NS
-	seq_put_decimal_ll(m, "\nNSpid:\t", nr);
-	if (nr > 0) {
-		int i;
-
-		/* If nr is non-zero it means that 'pid' is valid and that
-		 * ns, i.e. the pid namespace associated with the procfs
-		 * instance, is in the pid namespace hierarchy of pid.
-		 * Start at one below the already printed level.
-		 */
-		for (i = ns->level + 1; i <= pid->level; i++)
-			seq_put_decimal_ll(m, "\t", pid->numbers[i].nr);
-	}
-#endif
-	seq_putc(m, '\n');
-}
-#endif
-
-/*
- * Poll support for process exit notification.
- */
-static __poll_t pidfd_poll(struct file *file, struct poll_table_struct *pts)
-{
-	struct pid *pid = file->private_data;
-	__poll_t poll_flags = 0;
-
-	poll_wait(file, &pid->wait_pidfd, pts);
-
-	/*
-	 * Inform pollers only when the whole thread group exits.
-	 * If the thread group leader exits before all other threads in the
-	 * group, then poll(2) should block, similar to the wait(2) family.
-	 */
-	if (thread_group_exited(pid))
-		poll_flags = EPOLLIN | EPOLLRDNORM;
-
-	return poll_flags;
-}
-
-const struct file_operations pidfd_fops = {
-	.release = pidfd_release,
-	.poll = pidfd_poll,
-#ifdef CONFIG_PROC_FS
-	.show_fdinfo = pidfd_show_fdinfo,
-#endif
-};
-
 /**
  * __pidfd_prepare - allocate a new pidfd_file and reserve a pidfd
  * @pid:   the struct pid for which to create a pidfd
@@ -2131,20 +2020,20 @@ static int __pidfd_prepare(struct pid *pid, unsigned int flags, struct file **re
 	int pidfd;
 	struct file *pidfd_file;
 
-	if (flags & ~(O_NONBLOCK | O_RDWR | O_CLOEXEC))
-		return -EINVAL;
-
-	pidfd = get_unused_fd_flags(O_RDWR | O_CLOEXEC);
+	pidfd = get_unused_fd_flags(O_CLOEXEC);
 	if (pidfd < 0)
 		return pidfd;
 
-	pidfd_file = anon_inode_getfile("[pidfd]", &pidfd_fops, pid,
-					flags | O_RDWR | O_CLOEXEC);
+	pidfd_file = pidfs_alloc_file(pid, flags | O_RDWR);
 	if (IS_ERR(pidfd_file)) {
 		put_unused_fd(pidfd);
 		return PTR_ERR(pidfd_file);
 	}
-	get_pid(pid); /* held by pidfd_file now */
+	/*
+	 * anon_inode_getfile() ignores everything outside of the
+	 * O_ACCMODE | O_NONBLOCK mask, set PIDFD_THREAD manually.
+	 */
+	pidfd_file->f_flags |= (flags & PIDFD_THREAD);
 	*ret = pidfd_file;
 	return pidfd;
 }
@@ -2158,7 +2047,8 @@ static int __pidfd_prepare(struct pid *pid, unsigned int flags, struct file **re
  * Allocate a new file that stashes @pid and reserve a new pidfd number in the
  * caller's file descriptor table. The pidfd is reserved but not installed yet.
  *
- * The helper verifies that @pid is used as a thread group leader.
+ * The helper verifies that @pid is still in use, without PIDFD_THREAD the
+ * task identified by @pid must be a thread-group leader.
  *
  * If this function returns successfully the caller is responsible to either
  * call fd_install() passing the returned pidfd and pidfd file as arguments in
@@ -2177,7 +2067,9 @@ static int __pidfd_prepare(struct pid *pid, unsigned int flags, struct file **re
  */
 int pidfd_prepare(struct pid *pid, unsigned int flags, struct file **ret)
 {
-	if (!pid || !pid_has_task(pid, PIDTYPE_TGID))
+	bool thread = flags & PIDFD_THREAD;
+
+	if (!pid || !pid_has_task(pid, thread ? PIDTYPE_PID : PIDTYPE_TGID))
 		return -EINVAL;
 
 	return __pidfd_prepare(pid, flags, ret);
@@ -2299,9 +2191,8 @@ __latent_entropy struct task_struct *copy_process(
 		/*
 		 * - CLONE_DETACHED is blocked so that we can potentially
 		 *   reuse it later for CLONE_PIDFD.
-		 * - CLONE_THREAD is blocked until someone really needs it.
 		 */
-		if (clone_flags & (CLONE_DETACHED | CLONE_THREAD))
+		if (clone_flags & CLONE_DETACHED)
 			return ERR_PTR(-EINVAL);
 	}
 
@@ -2524,8 +2415,10 @@ __latent_entropy struct task_struct *copy_process(
 	 * if the fd table isn't shared).
 	 */
 	if (clone_flags & CLONE_PIDFD) {
+		int flags = (clone_flags & CLONE_THREAD) ? PIDFD_THREAD : 0;
+
 		/* Note that no task has been attached to @pid yet. */
-		retval = __pidfd_prepare(pid, O_RDWR | O_CLOEXEC, &pidfile);
+		retval = __pidfd_prepare(pid, flags, &pidfile);
 		if (retval < 0)
 			goto bad_fork_free_pid;
 		pidfd = retval;
@@ -2876,8 +2769,8 @@ pid_t kernel_clone(struct kernel_clone_args *args)
 	 * here has the advantage that we don't need to have a separate helper
 	 * to check for legacy clone().
 	 */
-	if ((args->flags & CLONE_PIDFD) &&
-	    (args->flags & CLONE_PARENT_SETTID) &&
+	if ((clone_flags & CLONE_PIDFD) &&
+	    (clone_flags & CLONE_PARENT_SETTID) &&
 	    (args->pidfd == args->parent_tid))
 		return -EINVAL;
 

kernel/nsproxy.c

@@ -573,7 +573,7 @@ SYSCALL_DEFINE2(setns, int, fd, int, flags)
 	if (proc_ns_file(f.file))
 		err = validate_ns(&nsset, ns);
 	else
-		err = validate_nsset(&nsset, f.file->private_data);
+		err = validate_nsset(&nsset, pidfd_pid(f.file));
 	if (!err) {
 		commit_nsset(&nsset);
 		perf_event_namespaces(current);

kernel/pid.c

@@ -42,6 +42,7 @@
 #include <linux/sched/signal.h>
 #include <linux/sched/task.h>
 #include <linux/idr.h>
+#include <linux/pidfs.h>
 #include <net/sock.h>
 #include <uapi/linux/pidfd.h>
 
@@ -65,6 +66,13 @@ int pid_max = PID_MAX_DEFAULT;
 
 int pid_max_min = RESERVED_PIDS + 1;
 int pid_max_max = PID_MAX_LIMIT;
+#ifdef CONFIG_FS_PID
+/*
+ * Pseudo filesystems start inode numbering after one. We use Reserved
+ * PIDs as a natural offset.
+ */
+static u64 pidfs_ino = RESERVED_PIDS;
+#endif
 
 /*
  * PID-map pages start out as NULL, they get allocated upon
@@ -272,6 +280,10 @@ struct pid *alloc_pid(struct pid_namespace *ns, pid_t *set_tid,
 	spin_lock_irq(&pidmap_lock);
 	if (!(ns->pid_allocated & PIDNS_ADDING))
 		goto out_unlock;
+#ifdef CONFIG_FS_PID
+	pid->stashed = NULL;
+	pid->ino = ++pidfs_ino;
+#endif
 	for ( ; upid >= pid->numbers; --upid) {
 		/* Make the PID visible to find_pid_ns. */
 		idr_replace(&upid->ns->idr, pid, upid->nr);
@@ -349,6 +361,11 @@ static void __change_pid(struct task_struct *task, enum pid_type type,
 	hlist_del_rcu(&task->pid_links[type]);
 	*pid_ptr = new;
 
+	if (type == PIDTYPE_PID) {
+		WARN_ON_ONCE(pid_has_task(pid, PIDTYPE_PID));
+		wake_up_all(&pid->wait_pidfd);
+	}
+
 	for (tmp = PIDTYPE_MAX; --tmp >= 0; )
 		if (pid_has_task(pid, tmp))
 			return;
@@ -391,8 +408,7 @@ void exchange_tids(struct task_struct *left, struct task_struct *right)
 void transfer_pid(struct task_struct *old, struct task_struct *new,
 			   enum pid_type type)
 {
-	if (type == PIDTYPE_PID)
-		new->thread_pid = old->thread_pid;
+	WARN_ON_ONCE(type == PIDTYPE_PID);
 	hlist_replace_rcu(&old->pid_links[type], &new->pid_links[type]);
 }
 
@@ -552,11 +568,6 @@ struct pid *pidfd_get_pid(unsigned int fd, unsigned int *flags)
  * Return the task associated with @pidfd. The function takes a reference on
  * the returned task. The caller is responsible for releasing that reference.
  *
- * Currently, the process identified by @pidfd is always a thread-group leader.
- * This restriction currently exists for all aspects of pidfds including pidfd
- * creation (CLONE_PIDFD cannot be used with CLONE_THREAD) and pidfd polling
- * (only supports thread group leaders).
- *
  * Return: On success, the task_struct associated with the pidfd.
  *	   On error, a negative errno number will be returned.
  */
@@ -595,7 +606,7 @@ struct task_struct *pidfd_get_task(int pidfd, unsigned int *flags)
  * Return: On success, a cloexec pidfd is returned.
  *         On error, a negative errno number will be returned.
  */
-int pidfd_create(struct pid *pid, unsigned int flags)
+static int pidfd_create(struct pid *pid, unsigned int flags)
 {
 	int pidfd;
 	struct file *pidfd_file;
@@ -615,11 +626,8 @@ int pidfd_create(struct pid *pid, unsigned int flags)
  * @flags: flags to pass
  *
  * This creates a new pid file descriptor with the O_CLOEXEC flag set for
- * the process identified by @pid. Currently, the process identified by
- * @pid must be a thread-group leader. This restriction currently exists
- * for all aspects of pidfds including pidfd creation (CLONE_PIDFD cannot
- * be used with CLONE_THREAD) and pidfd polling (only supports thread group
- * leaders).
+ * the task identified by @pid. Without PIDFD_THREAD flag the target task
+ * must be a thread-group leader.
  *
  * Return: On success, a cloexec pidfd is returned.
  *         On error, a negative errno number will be returned.
@@ -629,7 +637,7 @@ SYSCALL_DEFINE2(pidfd_open, pid_t, pid, unsigned int, flags)
 	int fd;
 	struct pid *p;
 
-	if (flags & ~PIDFD_NONBLOCK)
+	if (flags & ~(PIDFD_NONBLOCK | PIDFD_THREAD))
 		return -EINVAL;
 
 	if (pid <= 0)
@@ -682,7 +690,26 @@ static struct file *__pidfd_fget(struct task_struct *task, int fd)
 
 	up_read(&task->signal->exec_update_lock);
 
-	return file ?: ERR_PTR(-EBADF);
+	if (!file) {
+		/*
+		 * It is possible that the target thread is exiting; it can be
+		 * either:
+		 * 1. before exit_signals(), which gives a real fd
+		 * 2. before exit_files() takes the task_lock() gives a real fd
+		 * 3. after exit_files() releases task_lock(), ->files is NULL;
+		 *    this has PF_EXITING, since it was set in exit_signals(),
+		 *    __pidfd_fget() returns EBADF.
+		 * In case 3 we get EBADF, but that really means ESRCH, since
+		 * the task is currently exiting and has freed its files
+		 * struct, so we fix it up.
+		 */
+		if (task->flags & PF_EXITING)
+			file = ERR_PTR(-ESRCH);
+		else
+			file = ERR_PTR(-EBADF);
+	}
+
+	return file;
 }
 
 static int pidfd_getfd(struct pid *pid, int fd)

kernel/signal.c

@@ -47,6 +47,7 @@
 #include <linux/cgroup.h>
 #include <linux/audit.h>
 #include <linux/sysctl.h>
+#include <uapi/linux/pidfd.h>
 
 #define CREATE_TRACE_POINTS
 #include <trace/events/signal.h>
@@ -1436,7 +1437,8 @@ void lockdep_assert_task_sighand_held(struct task_struct *task)
 #endif
 
 /*
- * send signal info to all the members of a group
+ * send signal info to all the members of a thread group or to the
+ * individual thread if type == PIDTYPE_PID.
  */
 int group_send_sig_info(int sig, struct kernel_siginfo *info,
 			struct task_struct *p, enum pid_type type)
@@ -1478,7 +1480,8 @@ int __kill_pgrp_info(int sig, struct kernel_siginfo *info, struct pid *pgrp)
 	return ret;
 }
 
-int kill_pid_info(int sig, struct kernel_siginfo *info, struct pid *pid)
+static int kill_pid_info_type(int sig, struct kernel_siginfo *info,
+				struct pid *pid, enum pid_type type)
 {
 	int error = -ESRCH;
 	struct task_struct *p;
@@ -1487,11 +1490,10 @@ int kill_pid_info(int sig, struct kernel_siginfo *info, struct pid *pid)
 		rcu_read_lock();
 		p = pid_task(pid, PIDTYPE_PID);
 		if (p)
-			error = group_send_sig_info(sig, info, p, PIDTYPE_TGID);
+			error = group_send_sig_info(sig, info, p, type);
 		rcu_read_unlock();
 		if (likely(!p || error != -ESRCH))
 			return error;
-
 		/*
 		 * The task was unhashed in between, try again.  If it
 		 * is dead, pid_task() will return NULL, if we race with
@@ -1500,6 +1502,11 @@ int kill_pid_info(int sig, struct kernel_siginfo *info, struct pid *pid)
 	}
 }
 
+int kill_pid_info(int sig, struct kernel_siginfo *info, struct pid *pid)
+{
+	return kill_pid_info_type(sig, info, pid, PIDTYPE_TGID);
+}
+
 static int kill_proc_info(int sig, struct kernel_siginfo *info, pid_t pid)
 {
 	int error;
@@ -1898,16 +1905,19 @@ int send_sig_fault_trapno(int sig, int code, void __user *addr, int trapno,
 	return send_sig_info(info.si_signo, &info, t);
 }
 
-int kill_pgrp(struct pid *pid, int sig, int priv)
+static int kill_pgrp_info(int sig, struct kernel_siginfo *info, struct pid *pgrp)
 {
 	int ret;
-
 	read_lock(&tasklist_lock);
-	ret = __kill_pgrp_info(sig, __si_special(priv), pid);
+	ret = __kill_pgrp_info(sig, info, pgrp);
 	read_unlock(&tasklist_lock);
-
 	return ret;
 }
+
+int kill_pgrp(struct pid *pid, int sig, int priv)
+{
+	return kill_pgrp_info(sig, __si_special(priv), pid);
+}
 EXPORT_SYMBOL(kill_pgrp);
 
 int kill_pid(struct pid *pid, int sig, int priv)
@@ -2019,13 +2029,14 @@ int send_sigqueue(struct sigqueue *q, struct pid *pid, enum pid_type type)
 	return ret;
 }
 
-static void do_notify_pidfd(struct task_struct *task)
+void do_notify_pidfd(struct task_struct *task)
 {
-	struct pid *pid;
+	struct pid *pid = task_pid(task);
 
 	WARN_ON(task->exit_state == 0);
-	pid = task_pid(task);
-	wake_up_all(&pid->wait_pidfd);
+
+	__wake_up(&pid->wait_pidfd, TASK_NORMAL, 0,
+			poll_to_key(EPOLLIN | EPOLLRDNORM));
 }
 
 /*
@@ -2050,9 +2061,12 @@ bool do_notify_parent(struct task_struct *tsk, int sig)
 
 	WARN_ON_ONCE(!tsk->ptrace &&
 	       (tsk->group_leader != tsk || !thread_group_empty(tsk)));
-
-	/* Wake up all pidfd waiters */
-	do_notify_pidfd(tsk);
+	/*
+	 * tsk is a group leader and has no threads, wake up the
+	 * non-PIDFD_THREAD waiters.
+	 */
+	if (thread_group_empty(tsk))
+		do_notify_pidfd(tsk);
 
 	if (sig != SIGCHLD) {
 		/*
@@ -3789,12 +3803,13 @@ COMPAT_SYSCALL_DEFINE4(rt_sigtimedwait_time32, compat_sigset_t __user *, uthese,
 #endif
 #endif
 
-static inline void prepare_kill_siginfo(int sig, struct kernel_siginfo *info)
+static void prepare_kill_siginfo(int sig, struct kernel_siginfo *info,
+				 enum pid_type type)
 {
 	clear_siginfo(info);
 	info->si_signo = sig;
 	info->si_errno = 0;
-	info->si_code = SI_USER;
+	info->si_code = (type == PIDTYPE_PID) ? SI_TKILL : SI_USER;
 	info->si_pid = task_tgid_vnr(current);
 	info->si_uid = from_kuid_munged(current_user_ns(), current_uid());
 }
@@ -3808,7 +3823,7 @@ SYSCALL_DEFINE2(kill, pid_t, pid, int, sig)
 {
 	struct kernel_siginfo info;
 
-	prepare_kill_siginfo(sig, &info);
+	prepare_kill_siginfo(sig, &info, PIDTYPE_TGID);
 
 	return kill_something_info(sig, &info, pid);
 }
@@ -3861,6 +3876,10 @@ static struct pid *pidfd_to_pid(const struct file *file)
 	return tgid_pidfd_to_pid(file);
 }
 
+#define PIDFD_SEND_SIGNAL_FLAGS                            \
+	(PIDFD_SIGNAL_THREAD | PIDFD_SIGNAL_THREAD_GROUP | \
+	 PIDFD_SIGNAL_PROCESS_GROUP)
+
 /**
  * sys_pidfd_send_signal - Signal a process through a pidfd
  * @pidfd:  file descriptor of the process
@@ -3868,14 +3887,10 @@ static struct pid *pidfd_to_pid(const struct file *file)
  * @info:   signal info
  * @flags:  future flags
  *
- * The syscall currently only signals via PIDTYPE_PID which covers
- * kill(<positive-pid>, <signal>. It does not signal threads or process
- * groups.
- * In order to extend the syscall to threads and process groups the @flags
- * argument should be used. In essence, the @flags argument will determine
- * what is signaled and not the file descriptor itself. Put in other words,
- * grouping is a property of the flags argument not a property of the file
- * descriptor.
+ * Send the signal to the thread group or to the individual thread depending
+ * on PIDFD_THREAD.
+ * In the future extension to @flags may be used to override the default scope
+ * of @pidfd.
  *
  * Return: 0 on success, negative errno on failure
  */
@@ -3886,9 +3901,14 @@ SYSCALL_DEFINE4(pidfd_send_signal, int, pidfd, int, sig,
 	struct fd f;
 	struct pid *pid;
 	kernel_siginfo_t kinfo;
+	enum pid_type type;
 
 	/* Enforce flags be set to 0 until we add an extension. */
-	if (flags)
+	if (flags & ~PIDFD_SEND_SIGNAL_FLAGS)
+		return -EINVAL;
+
+	/* Ensure that only a single signal scope determining flag is set. */
+	if (hweight32(flags & PIDFD_SEND_SIGNAL_FLAGS) > 1)
 		return -EINVAL;
 
 	f = fdget(pidfd);
@@ -3906,6 +3926,25 @@ SYSCALL_DEFINE4(pidfd_send_signal, int, pidfd, int, sig,
 	if (!access_pidfd_pidns(pid))
 		goto err;
 
+	switch (flags) {
+	case 0:
+		/* Infer scope from the type of pidfd. */
+		if (f.file->f_flags & PIDFD_THREAD)
+			type = PIDTYPE_PID;
+		else
+			type = PIDTYPE_TGID;
+		break;
+	case PIDFD_SIGNAL_THREAD:
+		type = PIDTYPE_PID;
+		break;
+	case PIDFD_SIGNAL_THREAD_GROUP:
+		type = PIDTYPE_TGID;
+		break;
+	case PIDFD_SIGNAL_PROCESS_GROUP:
+		type = PIDTYPE_PGID;
+		break;
+	}
+
 	if (info) {
 		ret = copy_siginfo_from_user_any(&kinfo, info);
 		if (unlikely(ret))
@@ -3917,15 +3956,17 @@ SYSCALL_DEFINE4(pidfd_send_signal, int, pidfd, int, sig,
 
 		/* Only allow sending arbitrary signals to yourself. */
 		ret = -EPERM;
-		if ((task_pid(current) != pid) &&
+		if ((task_pid(current) != pid || type > PIDTYPE_TGID) &&
 		    (kinfo.si_code >= 0 || kinfo.si_code == SI_TKILL))
 			goto err;
 	} else {
-		prepare_kill_siginfo(sig, &kinfo);
+		prepare_kill_siginfo(sig, &kinfo, type);
 	}
 
-	ret = kill_pid_info(sig, &kinfo, pid);
-
+	if (type == PIDTYPE_PGID)
+		ret = kill_pgrp_info(sig, &kinfo, pid);
+	else
+		ret = kill_pid_info_type(sig, &kinfo, pid, type);
 err:
 	fdput(f);
 	return ret;
@@ -3965,12 +4006,7 @@ static int do_tkill(pid_t tgid, pid_t pid, int sig)
 {
 	struct kernel_siginfo info;
 
-	clear_siginfo(&info);
-	info.si_signo = sig;
-	info.si_errno = 0;
-	info.si_code = SI_TKILL;
-	info.si_pid = task_tgid_vnr(current);
-	info.si_uid = from_kuid_munged(current_user_ns(), current_uid());
+	prepare_kill_siginfo(sig, &info, PIDTYPE_PID);
 
 	return do_send_specific(tgid, pid, sig, &info);
 }

tools/testing/selftests/pidfd/pidfd_getfd_test.c

@@ -5,6 +5,7 @@
 #include <fcntl.h>
 #include <limits.h>
 #include <linux/types.h>
+#include <poll.h>
 #include <sched.h>
 #include <signal.h>
 #include <stdio.h>
@@ -129,6 +130,7 @@ FIXTURE(child)
 	 * When it is closed, the child will exit.
 	 */
 	int sk;
+	bool ignore_child_result;
 };
 
 FIXTURE_SETUP(child)
@@ -165,10 +167,14 @@ FIXTURE_SETUP(child)
 
 FIXTURE_TEARDOWN(child)
 {
+	int ret;
+
 	EXPECT_EQ(0, close(self->pidfd));
 	EXPECT_EQ(0, close(self->sk));
 
-	EXPECT_EQ(0, wait_for_pid(self->pid));
+	ret = wait_for_pid(self->pid);
+	if (!self->ignore_child_result)
+		EXPECT_EQ(0, ret);
 }
 
 TEST_F(child, disable_ptrace)
@@ -235,6 +241,29 @@ TEST(flags_set)
 	EXPECT_EQ(errno, EINVAL);
 }
 
+TEST_F(child, no_strange_EBADF)
+{
+	struct pollfd fds;
+
+	self->ignore_child_result = true;
+
+	fds.fd = self->pidfd;
+	fds.events = POLLIN;
+
+	ASSERT_EQ(kill(self->pid, SIGKILL), 0);
+	ASSERT_EQ(poll(&fds, 1, 5000), 1);
+
+	/*
+	 * It used to be that pidfd_getfd() could race with the exiting thread
+	 * between exit_files() and release_task(), and get a non-null task
+	 * with a NULL files struct, and you'd get EBADF, which was slightly
+	 * confusing.
+	 */
+	errno = 0;
+	EXPECT_EQ(sys_pidfd_getfd(self->pidfd, self->remote_fd, 0), -1);
+	EXPECT_EQ(errno, ESRCH);
+}
+
 #if __NR_pidfd_getfd == -1
 int main(void)
 {