[v2,1/2] Add polling support to pidfd
diff mbox series

Message ID 20190430162154.61314-1-joel@joelfernandes.org
State New
Headers show
Series
  • [v2,1/2] Add polling support to pidfd
Related show

Commit Message

Joel Fernandes April 30, 2019, 4:21 p.m. UTC
Android low memory killer (LMK) needs to know when a process dies once
it is sent the kill signal. It does so by checking for the existence of
/proc/pid which is both racy and slow. For example, if a PID is reused
between when LMK sends a kill signal and checks for existence of the
PID, since the wrong PID is now possibly checked for existence.

This patch adds polling support to pidfd. Using the polling support, LMK
will be able to get notified when a process exists in race-free and fast
way, and allows the LMK to do other things (such as by polling on other
fds) while awaiting the process being killed to die.

For notification to polling processes, we follow the same existing
mechanism in the kernel used when the parent of the task group is to be
notified of a child's death (do_notify_parent).  This is precisely when
the tasks waiting on a poll of pidfd are also awakened in this patch.

We have decided to include the waitqueue in struct pid for the following
reasons:
1. The wait queue has to survive for the lifetime of the poll. Including
it in task_struct would not be option in this case because the task can
be reaped and destroyed before the poll returns.

2. By including the struct pid for the waitqueue means that during
de_thread(), the new thread group leader automatically gets the new
waitqueue/pid even though its task_struct is different.

Appropriate test cases are added in the second patch to provide coverage
of all the cases the patch is handling.

Andy had a similar patch [1] in the past which was a good reference
however this patch tries to handle different situations properly related
to thread group existence, and how/where it notifies. And also solves
other bugs (waitqueue lifetime).  Daniel had a similar patch [2]
recently which this patch supercedes.

[1] https://lore.kernel.org/patchwork/patch/345098/
[2] https://lore.kernel.org/lkml/20181029175322.189042-1-dancol@google.com/

Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Daniel Colascione <dancol@google.com>
Cc: Christian Brauner <christian@brauner.io>
Cc: Jann Horn <jannh@google.com>
Cc: Tim Murray <timmurray@google.com>
Cc: Jonathan Kowalski <bl0pbl33p@gmail.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Kees Cook <keescook@chromium.org>
Cc: David Howells <dhowells@redhat.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: kernel-team@android.com
(Oleg improved the code by showing how to avoid tasklist_lock)
Suggested-by: Oleg Nesterov <oleg@redhat.com>
Co-developed-by: Daniel Colascione <dancol@google.com>
Signed-off-by: Daniel Colascione <dancol@google.com>
Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org>

---

v1 -> v2:
* Restructure poll code to avoid tasklist_lock (Oleg)
* use task_pid instead of get_pid_task in notify_pidfd (Oleg)
* Added comments to code, commit message nits (Christian)
* Test case nits/improvements (Christian)

RFC -> v1:
* Based on CLONE_PIDFD patches: https://lwn.net/Articles/786244/
* Updated selftests.
* Renamed poll wake function to do_notify_pidfd.
* Removed depending on EXIT flags
* Removed POLLERR flag since semantics are controversial and
  we don't have usecases for it right now (later we can add if there's
  a need for it).

 include/linux/pid.h |  3 +++
 kernel/fork.c       | 29 +++++++++++++++++++++++++++++
 kernel/pid.c        |  2 ++
 kernel/signal.c     | 11 +++++++++++
 4 files changed, 45 insertions(+)

Comments

Christian Brauner April 30, 2019, 7:07 p.m. UTC | #1
On Tue, Apr 30, 2019 at 12:21:53PM -0400, Joel Fernandes (Google) wrote:
> Android low memory killer (LMK) needs to know when a process dies once
> it is sent the kill signal. It does so by checking for the existence of
> /proc/pid which is both racy and slow. For example, if a PID is reused
> between when LMK sends a kill signal and checks for existence of the
> PID, since the wrong PID is now possibly checked for existence.
> 
> This patch adds polling support to pidfd. Using the polling support, LMK
> will be able to get notified when a process exists in race-free and fast
> way, and allows the LMK to do other things (such as by polling on other
> fds) while awaiting the process being killed to die.
> 
> For notification to polling processes, we follow the same existing
> mechanism in the kernel used when the parent of the task group is to be
> notified of a child's death (do_notify_parent).  This is precisely when
> the tasks waiting on a poll of pidfd are also awakened in this patch.
> 
> We have decided to include the waitqueue in struct pid for the following
> reasons:
> 1. The wait queue has to survive for the lifetime of the poll. Including
> it in task_struct would not be option in this case because the task can
> be reaped and destroyed before the poll returns.
> 
> 2. By including the struct pid for the waitqueue means that during
> de_thread(), the new thread group leader automatically gets the new
> waitqueue/pid even though its task_struct is different.
> 
> Appropriate test cases are added in the second patch to provide coverage
> of all the cases the patch is handling.
> 
> Andy had a similar patch [1] in the past which was a good reference
> however this patch tries to handle different situations properly related
> to thread group existence, and how/where it notifies. And also solves
> other bugs (waitqueue lifetime).  Daniel had a similar patch [2]
> recently which this patch supercedes.
> 
> [1] https://lore.kernel.org/patchwork/patch/345098/
> [2] https://lore.kernel.org/lkml/20181029175322.189042-1-dancol@google.com/
> 
> Cc: Andy Lutomirski <luto@amacapital.net>
> Cc: Steven Rostedt <rostedt@goodmis.org>
> Cc: Daniel Colascione <dancol@google.com>
> Cc: Christian Brauner <christian@brauner.io>
> Cc: Jann Horn <jannh@google.com>
> Cc: Tim Murray <timmurray@google.com>
> Cc: Jonathan Kowalski <bl0pbl33p@gmail.com>
> Cc: Linus Torvalds <torvalds@linux-foundation.org>
> Cc: Al Viro <viro@zeniv.linux.org.uk>
> Cc: Kees Cook <keescook@chromium.org>
> Cc: David Howells <dhowells@redhat.com>
> Cc: Oleg Nesterov <oleg@redhat.com>
> Cc: kernel-team@android.com
> (Oleg improved the code by showing how to avoid tasklist_lock)
> Suggested-by: Oleg Nesterov <oleg@redhat.com>
> Co-developed-by: Daniel Colascione <dancol@google.com>
> Signed-off-by: Daniel Colascione <dancol@google.com>
> Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org>

This looks good to me. Once Oleg has given his Ack/Review I'm going to
move this into pidfd for-next and as mentioned before schedule it for
5.3 after CLONE_PIDFD has been merged.

Reviewed-by: Christian Brauner <christian@brauner.io>

> 
> ---
> 
> v1 -> v2:
> * Restructure poll code to avoid tasklist_lock (Oleg)
> * use task_pid instead of get_pid_task in notify_pidfd (Oleg)
> * Added comments to code, commit message nits (Christian)
> * Test case nits/improvements (Christian)
> 
> RFC -> v1:
> * Based on CLONE_PIDFD patches: https://lwn.net/Articles/786244/
> * Updated selftests.
> * Renamed poll wake function to do_notify_pidfd.
> * Removed depending on EXIT flags
> * Removed POLLERR flag since semantics are controversial and
>   we don't have usecases for it right now (later we can add if there's
>   a need for it).
> 
>  include/linux/pid.h |  3 +++
>  kernel/fork.c       | 29 +++++++++++++++++++++++++++++
>  kernel/pid.c        |  2 ++
>  kernel/signal.c     | 11 +++++++++++
>  4 files changed, 45 insertions(+)
> 
> diff --git a/include/linux/pid.h b/include/linux/pid.h
> index 3c8ef5a199ca..1484db6ca8d1 100644
> --- a/include/linux/pid.h
> +++ b/include/linux/pid.h
> @@ -3,6 +3,7 @@
>  #define _LINUX_PID_H
>  
>  #include <linux/rculist.h>
> +#include <linux/wait.h>
>  
>  enum pid_type
>  {
> @@ -60,6 +61,8 @@ struct pid
>  	unsigned int level;
>  	/* lists of tasks that use this pid */
>  	struct hlist_head tasks[PIDTYPE_MAX];
> +	/* wait queue for pidfd notifications */
> +	wait_queue_head_t wait_pidfd;
>  	struct rcu_head rcu;
>  	struct upid numbers[1];
>  };
> diff --git a/kernel/fork.c b/kernel/fork.c
> index 5525837ed80e..721f8c9d2921 100644
> --- a/kernel/fork.c
> +++ b/kernel/fork.c
> @@ -1685,8 +1685,37 @@ static void pidfd_show_fdinfo(struct seq_file *m, struct file *f)
>  }
>  #endif
>  
> +/*
> + * Poll support for process exit notification.
> + */
> +static unsigned int pidfd_poll(struct file *file, struct poll_table_struct *pts)
> +{
> +	struct task_struct *task;
> +	struct pid *pid = file->private_data;
> +	int poll_flags = 0;
> +
> +	poll_wait(file, &pid->wait_pidfd, pts);
> +
> +	rcu_read_lock();
> +	task = pid_task(pid, PIDTYPE_PID);
> +	WARN_ON_ONCE(task && !thread_group_leader(task));
> +
> +	/*
> +	 * Inform pollers only when the whole thread group exits, if thread
> +	 * group leader exits before all other threads in the group, then
> +	 * poll(2) should block, similar to the wait(2) family.
> +	 */
> +	if (!task || (task->exit_state && thread_group_empty(task)))
> +		poll_flags = POLLIN | POLLRDNORM;
> +	rcu_read_unlock();
> +
> +	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
> diff --git a/kernel/pid.c b/kernel/pid.c
> index 20881598bdfa..5c90c239242f 100644
> --- a/kernel/pid.c
> +++ b/kernel/pid.c
> @@ -214,6 +214,8 @@ struct pid *alloc_pid(struct pid_namespace *ns)
>  	for (type = 0; type < PIDTYPE_MAX; ++type)
>  		INIT_HLIST_HEAD(&pid->tasks[type]);
>  
> +	init_waitqueue_head(&pid->wait_pidfd);
> +
>  	upid = pid->numbers + ns->level;
>  	spin_lock_irq(&pidmap_lock);
>  	if (!(ns->pid_allocated & PIDNS_ADDING))
> diff --git a/kernel/signal.c b/kernel/signal.c
> index 1581140f2d99..a17fff073c3d 100644
> --- a/kernel/signal.c
> +++ b/kernel/signal.c
> @@ -1800,6 +1800,14 @@ int send_sigqueue(struct sigqueue *q, struct pid *pid, enum pid_type type)
>  	return ret;
>  }
>  
> +static void do_notify_pidfd(struct task_struct *task)
> +{
> +	struct pid *pid;
> +
> +	pid = task_pid(task);
> +	wake_up_all(&pid->wait_pidfd);
> +}
> +
>  /*
>   * Let a parent know about the death of a child.
>   * For a stopped/continued status change, use do_notify_parent_cldstop instead.
> @@ -1823,6 +1831,9 @@ bool do_notify_parent(struct task_struct *tsk, int sig)
>  	BUG_ON(!tsk->ptrace &&
>  	       (tsk->group_leader != tsk || !thread_group_empty(tsk)));
>  
> +	/* Wake up all pidfd waiters */
> +	do_notify_pidfd(tsk);
> +
>  	if (sig != SIGCHLD) {
>  		/*
>  		 * This is only possible if parent == real_parent.
> -- 
> 2.21.0.593.g511ec345e18-goog
Oleg Nesterov May 1, 2019, 3:13 p.m. UTC | #2
On 04/30, Joel Fernandes (Google) wrote:
>
> +static unsigned int pidfd_poll(struct file *file, struct poll_table_struct *pts)
> +{
> +	struct task_struct *task;
> +	struct pid *pid = file->private_data;
> +	int poll_flags = 0;
> +
> +	poll_wait(file, &pid->wait_pidfd, pts);
> +
> +	rcu_read_lock();
> +	task = pid_task(pid, PIDTYPE_PID);
> +	WARN_ON_ONCE(task && !thread_group_leader(task));
                             ^^^^^^^^^^^^^^^^^^^^^^^^^^

Ah, this is not right, we can race with de_thread() which changes the leader,
in particular it does leader->exit_signal = -1 to indicate that this thread is
no longer a group leader, but pid_task() can return the old leader.

We are going to check thread_group_empty() below, it won't be true in this case,
so this race should not make any harm.

Just remove this WARN_ON(). We can't use has_group_leader_pid(), it can return
false if pid_task() returns the new leader.

Otherwise I see no problems.

Oleg.
Christian Brauner May 2, 2019, 3:13 p.m. UTC | #3
On Wed, May 01, 2019 at 05:13:12PM +0200, Oleg Nesterov wrote:
> On 04/30, Joel Fernandes (Google) wrote:
> >
> > +static unsigned int pidfd_poll(struct file *file, struct poll_table_struct *pts)
> > +{
> > +	struct task_struct *task;
> > +	struct pid *pid = file->private_data;
> > +	int poll_flags = 0;
> > +
> > +	poll_wait(file, &pid->wait_pidfd, pts);
> > +
> > +	rcu_read_lock();
> > +	task = pid_task(pid, PIDTYPE_PID);
> > +	WARN_ON_ONCE(task && !thread_group_leader(task));
>                              ^^^^^^^^^^^^^^^^^^^^^^^^^^
> 
> Ah, this is not right, we can race with de_thread() which changes the leader,
> in particular it does leader->exit_signal = -1 to indicate that this thread is
> no longer a group leader, but pid_task() can return the old leader.
> 
> We are going to check thread_group_empty() below, it won't be true in this case,
> so this race should not make any harm.
> 
> Just remove this WARN_ON(). We can't use has_group_leader_pid(), it can return
> false if pid_task() returns the new leader.
> 
> Otherwise I see no problems.

I'll remove the WARN_ON() check when applying this. Can I get your
Acked/Review, Oleg?

Christian
Oleg Nesterov May 2, 2019, 4:02 p.m. UTC | #4
On 05/02, Christian Brauner wrote:
>
> On Wed, May 01, 2019 at 05:13:12PM +0200, Oleg Nesterov wrote:
> >
> > Otherwise I see no problems.
>
> I'll remove the WARN_ON() check when applying this. Can I get your
> Acked/Review, Oleg?

Yes, feel free to add

	Reviewed-by: Oleg Nesterov <oleg@redhat.com>


Hmm. Somehow I didn't read the changelog before, I just noticed
Suggested-by: Oleg Nesterov <oleg@redhat.com>

Please remove ;) Thanks Joel, I appreciate it, but it is not my idea.

Oleg.
Joel Fernandes May 2, 2019, 7:14 p.m. UTC | #5
On Thu, May 02, 2019 at 06:02:48PM +0200, Oleg Nesterov wrote:
> On 05/02, Christian Brauner wrote:
> >
> > On Wed, May 01, 2019 at 05:13:12PM +0200, Oleg Nesterov wrote:
> > >
> > > Otherwise I see no problems.
> >
> > I'll remove the WARN_ON() check when applying this. Can I get your
> > Acked/Review, Oleg?

Oh, ok. Good point about the de_thread race. Agreed with you.

> Yes, feel free to add
> 
> 	Reviewed-by: Oleg Nesterov <oleg@redhat.com>
> 
> Hmm. Somehow I didn't read the changelog before, I just noticed
> Suggested-by: Oleg Nesterov <oleg@redhat.com>
> Please remove ;) Thanks Joel, I appreciate it, but it is not my idea.

Ok no problem. You have been very helpful so thank you for that!

Also thanks Christian for removing the warning and adding Oleg's Reviewed-by.

 - Joel
Christian Brauner May 2, 2019, 7:35 p.m. UTC | #6
On Thu, May 02, 2019 at 03:14:37PM -0400, Joel Fernandes wrote:
> On Thu, May 02, 2019 at 06:02:48PM +0200, Oleg Nesterov wrote:
> > On 05/02, Christian Brauner wrote:
> > >
> > > On Wed, May 01, 2019 at 05:13:12PM +0200, Oleg Nesterov wrote:
> > > >
> > > > Otherwise I see no problems.
> > >
> > > I'll remove the WARN_ON() check when applying this. Can I get your
> > > Acked/Review, Oleg?
> 
> Oh, ok. Good point about the de_thread race. Agreed with you.
> 
> > Yes, feel free to add
> > 
> > 	Reviewed-by: Oleg Nesterov <oleg@redhat.com>
> > 
> > Hmm. Somehow I didn't read the changelog before, I just noticed
> > Suggested-by: Oleg Nesterov <oleg@redhat.com>
> > Please remove ;) Thanks Joel, I appreciate it, but it is not my idea.
> 
> Ok no problem. You have been very helpful so thank you for that!

Yep, big thank you, Oleg! :)

Christian
Christian Brauner May 14, 2019, 3:16 p.m. UTC | #7
On Thu, May 02, 2019 at 06:02:48PM +0200, Oleg Nesterov wrote:
> On 05/02, Christian Brauner wrote:
> >
> > On Wed, May 01, 2019 at 05:13:12PM +0200, Oleg Nesterov wrote:
> > >
> > > Otherwise I see no problems.
> >
> > I'll remove the WARN_ON() check when applying this. Can I get your
> > Acked/Review, Oleg?
> 
> Yes, feel free to add
> 
> 	Reviewed-by: Oleg Nesterov <oleg@redhat.com>

Applied to 

https://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux.git/log/?h=for-next

and targeted for the 5.3 merge window.

Thank you all!
Christian

Patch
diff mbox series

diff --git a/include/linux/pid.h b/include/linux/pid.h
index 3c8ef5a199ca..1484db6ca8d1 100644
--- a/include/linux/pid.h
+++ b/include/linux/pid.h
@@ -3,6 +3,7 @@ 
 #define _LINUX_PID_H
 
 #include <linux/rculist.h>
+#include <linux/wait.h>
 
 enum pid_type
 {
@@ -60,6 +61,8 @@  struct pid
 	unsigned int level;
 	/* lists of tasks that use this pid */
 	struct hlist_head tasks[PIDTYPE_MAX];
+	/* wait queue for pidfd notifications */
+	wait_queue_head_t wait_pidfd;
 	struct rcu_head rcu;
 	struct upid numbers[1];
 };
diff --git a/kernel/fork.c b/kernel/fork.c
index 5525837ed80e..721f8c9d2921 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -1685,8 +1685,37 @@  static void pidfd_show_fdinfo(struct seq_file *m, struct file *f)
 }
 #endif
 
+/*
+ * Poll support for process exit notification.
+ */
+static unsigned int pidfd_poll(struct file *file, struct poll_table_struct *pts)
+{
+	struct task_struct *task;
+	struct pid *pid = file->private_data;
+	int poll_flags = 0;
+
+	poll_wait(file, &pid->wait_pidfd, pts);
+
+	rcu_read_lock();
+	task = pid_task(pid, PIDTYPE_PID);
+	WARN_ON_ONCE(task && !thread_group_leader(task));
+
+	/*
+	 * Inform pollers only when the whole thread group exits, if thread
+	 * group leader exits before all other threads in the group, then
+	 * poll(2) should block, similar to the wait(2) family.
+	 */
+	if (!task || (task->exit_state && thread_group_empty(task)))
+		poll_flags = POLLIN | POLLRDNORM;
+	rcu_read_unlock();
+
+	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
diff --git a/kernel/pid.c b/kernel/pid.c
index 20881598bdfa..5c90c239242f 100644
--- a/kernel/pid.c
+++ b/kernel/pid.c
@@ -214,6 +214,8 @@  struct pid *alloc_pid(struct pid_namespace *ns)
 	for (type = 0; type < PIDTYPE_MAX; ++type)
 		INIT_HLIST_HEAD(&pid->tasks[type]);
 
+	init_waitqueue_head(&pid->wait_pidfd);
+
 	upid = pid->numbers + ns->level;
 	spin_lock_irq(&pidmap_lock);
 	if (!(ns->pid_allocated & PIDNS_ADDING))
diff --git a/kernel/signal.c b/kernel/signal.c
index 1581140f2d99..a17fff073c3d 100644
--- a/kernel/signal.c
+++ b/kernel/signal.c
@@ -1800,6 +1800,14 @@  int send_sigqueue(struct sigqueue *q, struct pid *pid, enum pid_type type)
 	return ret;
 }
 
+static void do_notify_pidfd(struct task_struct *task)
+{
+	struct pid *pid;
+
+	pid = task_pid(task);
+	wake_up_all(&pid->wait_pidfd);
+}
+
 /*
  * Let a parent know about the death of a child.
  * For a stopped/continued status change, use do_notify_parent_cldstop instead.
@@ -1823,6 +1831,9 @@  bool do_notify_parent(struct task_struct *tsk, int sig)
 	BUG_ON(!tsk->ptrace &&
 	       (tsk->group_leader != tsk || !thread_group_empty(tsk)));
 
+	/* Wake up all pidfd waiters */
+	do_notify_pidfd(tsk);
+
 	if (sig != SIGCHLD) {
 		/*
 		 * This is only possible if parent == real_parent.