| Message ID | 20190425190010.46489-1-joel@joelfernandes.org (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | [v1,1/2] Add polling support to pidfd | expand |
On Thu, Apr 25, 2019 at 03:00:09PM -0400, Joel Fernandes (Google) wrote: > pidfd are file descriptors referring to a process created with the > CLONE_PIDFD clone(2) flag. Android low memory killer (LMK) needs pidfd > polling support to replace code that currently checks for existence of > /proc/pid for knowing that a process that is signalled to be killed has > died, which is both racy and slow. The pidfd poll approach is race-free, > and also allows the LMK to do other things (such as by polling on other > fds) while awaiting the process being killed to die. Thanks for the patch! Ok, let me be a little bit anal. Please start the commit message with what this patch does and then add the justification why. You just say the "pidfd-poll" approach. You can probably assume that CLONE_PIDFD is available for this patch. So something like: "This patch makes pidfds pollable. Specifically, it allows listeners to be informed when the process the pidfd referes to exits. This patch only introduces the ability to poll thread-group leaders since pidfds currently can only reference those..." Then justify the use-case and then go into implementation details. That's usually how I would think about this: - Change the codebase to do X - Why do we need X - Are there any technical details worth mentioning in the commit message (- Are there any controversial points that people stumbled upon but that have been settled sufficiently.) > pidfd are file descriptors referring to a process created with the > CLONE_PIDFD clone(2) flag. Android low memory killer (LMK) needs pidfd > polling support to replace code that currently checks for existence of > /proc/pid for knowing that a process that is signalled to be killed has > died, which is both racy and slow. The pidfd poll approach is race-free, > and also allows the LMK to do other things (such as by polling on other > fds) while awaiting the process being killed to die. > > It prevents a situation where 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. > > In this patch, we follow the same existing mechanism in the kernel used > when the parent of the task group is to be notified (do_notify_parent). > This is when the tasks waiting on a poll of pidfd are also awakened. > > 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: luto@amacapital.net > Cc: rostedt@goodmis.org > Cc: dancol@google.com > Cc: sspatil@google.com > Cc: christian@brauner.io > Cc: jannh@google.com > Cc: surenb@google.com > Cc: timmurray@google.com > Cc: Jonathan Kowalski <bl0pbl33p@gmail.com> > Cc: torvalds@linux-foundation.org > Cc: kernel-team@android.com These should all be in the form: Cc: Firstname Lastname <email@address.com> There are people missing from the Cc that really should be there... Even though he usually doesn't respond that often, please Cc Al on this. If he responds it's usually rather important. Oleg has reviewed your RFC patch quite substantially and given valuable feedback and has an opinion on this thing and is best acquainted with the exit code. So please add him to the Cc of the commit message in the appropriate form and also add him to the Cc of the thread. Probably also want linux-api for good measure since a lot of people are subscribed that would care about pollable pidfds. I'd also add Kees since he had some interest in this work and David (Howells). > Co-developed-by: Daniel Colascione <dancol@google.com> Every CDB needs to give a SOB as well. > Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org> > > --- > > 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 | 33 +++++++++++++++++++++++++++++++++ > kernel/pid.c | 2 ++ > kernel/signal.c | 14 ++++++++++++++ > 4 files changed, 52 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..fb3b614f6456 100644 > --- a/kernel/fork.c > +++ b/kernel/fork.c > @@ -1685,8 +1685,41 @@ static void pidfd_show_fdinfo(struct seq_file *m, struct file *f) > } > #endif > > +static unsigned int pidfd_poll(struct file *file, struct poll_table_struct *pts) > +{ > + struct task_struct *task; > + struct pid *pid; > + int poll_flags = 0; > + > + /* > + * tasklist_lock must be held because to avoid racing with > + * changes in exit_state and wake up. Basically to avoid: > + * > + * P0: read exit_state = 0 > + * P1: write exit_state = EXIT_DEAD > + * P1: Do a wake up - wq is empty, so do nothing > + * P0: Queue for polling - wait forever. > + */ > + read_lock(&tasklist_lock); > + pid = file->private_data; > + task = pid_task(pid, PIDTYPE_PID); > + WARN_ON_ONCE(task && !thread_group_leader(task)); > + > + if (!task || (task->exit_state && thread_group_empty(task))) > + poll_flags = POLLIN | POLLRDNORM; So we block until the thread-group is empty? Hm, the thread-group leader remains in zombie state until all threads are gone. Should probably just be a short comment somewhere that callers are only informed about a whole thread-group exit and not about when the thread-group leader has actually exited. I would like the ability to extend this interface in the future to allow for actually reading data from the pidfd on EPOLLIN. POSIX specifies that POLLIN and POLLRDNORM are set even if the message to be read is zero. So one cheap way of doing this would probably be to do a 0 read/ioctl. That wouldn't hurt your very limited usecase and people could test whether the read returned non-0 data and if so they know this interface got extended. If we never extend it here it won't matter. > + > + if (!poll_flags) > + poll_wait(file, &pid->wait_pidfd, pts); > + > + read_unlock(&tasklist_lock); > + > + 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..16e7718316e5 100644 > --- a/kernel/signal.c > +++ b/kernel/signal.c > @@ -1800,6 +1800,17 @@ int send_sigqueue(struct sigqueue *q, struct pid *pid, enum pid_type type) > return ret; > } > > +static void do_notify_pidfd(struct task_struct *task) Maybe a short command that this helper can only be called when we know that task is a thread-group leader wouldn't hurt so there's no confusion later. > +{ > + struct pid *pid; > + > + lockdep_assert_held(&tasklist_lock); > + > + pid = get_task_pid(task, PIDTYPE_PID); > + wake_up_all(&pid->wait_pidfd); > + put_pid(pid); > +} > + > /* > * Let a parent know about the death of a child. > * For a stopped/continued status change, use do_notify_parent_cldstop instead. > @@ -1823,6 +1834,9 @@ 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
On Fri, Apr 26, 2019 at 12:24:04AM +0200, Christian Brauner wrote: > On Thu, Apr 25, 2019 at 03:00:09PM -0400, Joel Fernandes (Google) wrote: > > pidfd are file descriptors referring to a process created with the > > CLONE_PIDFD clone(2) flag. Android low memory killer (LMK) needs pidfd > > polling support to replace code that currently checks for existence of > > /proc/pid for knowing that a process that is signalled to be killed has > > died, which is both racy and slow. The pidfd poll approach is race-free, > > and also allows the LMK to do other things (such as by polling on other > > fds) while awaiting the process being killed to die. > > Thanks for the patch! > > Ok, let me be a little bit anal. > Please start the commit message with what this patch does and then add The subject title is "Add polling support to pidfd", but ok I should write a better commit message. > the justification why. You just say the "pidfd-poll" approach. You can > probably assume that CLONE_PIDFD is available for this patch. So > something like: > > "This patch makes pidfds pollable. Specifically, it allows listeners to > be informed when the process the pidfd referes to exits. This patch only > introduces the ability to poll thread-group leaders since pidfds > currently can only reference those..." > > Then justify the use-case and then go into implementation details. > That's usually how I would think about this: > - Change the codebase to do X > - Why do we need X > - Are there any technical details worth mentioning in the commit message > (- Are there any controversial points that people stumbled upon but that > have been settled sufficiently.) Generally the "how" in the patch should be in the code, but ok. I changed the first 3 paragraphs of the changelog to the following, is that better? : 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. > > pidfd are file descriptors referring to a process created with the > > CLONE_PIDFD clone(2) flag. Android low memory killer (LMK) needs pidfd > > polling support to replace code that currently checks for existence of > > /proc/pid for knowing that a process that is signalled to be killed has > > died, which is both racy and slow. The pidfd poll approach is race-free, > > and also allows the LMK to do other things (such as by polling on other > > fds) while awaiting the process being killed to die. > > > > > It prevents a situation where 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. > > > > In this patch, we follow the same existing mechanism in the kernel used > > when the parent of the task group is to be notified (do_notify_parent). > > This is when the tasks waiting on a poll of pidfd are also awakened. > > > > 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: luto@amacapital.net > > Cc: rostedt@goodmis.org > > Cc: dancol@google.com > > Cc: sspatil@google.com > > Cc: christian@brauner.io > > Cc: jannh@google.com > > Cc: surenb@google.com > > Cc: timmurray@google.com > > Cc: Jonathan Kowalski <bl0pbl33p@gmail.com> > > Cc: torvalds@linux-foundation.org > > Cc: kernel-team@android.com > > These should all be in the form: > > Cc: Firstname Lastname <email@address.com> If this bothers you too much, I can also just remove the CC list from the changelog here, and include it in my invocation of git-send-email instead.. but I have seen commits in the tree that don't follow this rule. > > There are people missing from the Cc that really should be there... If you look at the CC list of the email, people in the get_maintainer.pl script were also added. I did run get_maintainer.pl and checkpatch. But ok, I will add the folks you are suggesting as well. Thanks. > Even though he usually doesn't respond that often, please Cc Al on this. > If he responds it's usually rather important. No issues on that, but I am wondering if he should also be in MAINTAINERS file somewhere such that get_maintainer.pl does pick him up. I added him. > Oleg has reviewed your RFC patch quite substantially and given valuable > feedback and has an opinion on this thing and is best acquainted with > the exit code. So please add him to the Cc of the commit message in the > appropriate form and also add him to the Cc of the thread. Done. > Probably also want linux-api for good measure since a lot of people are > subscribed that would care about pollable pidfds. I'd also add Kees > since he had some interest in this work and David (Howells). Done, I added all of them and CC will go out to them next time. Thanks. > > > Co-developed-by: Daniel Colascione <dancol@google.com> > > Every CDB needs to give a SOB as well. Ok, done. thanks. > > > Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org> > > > > --- > > > > 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 | 33 +++++++++++++++++++++++++++++++++ > > kernel/pid.c | 2 ++ > > kernel/signal.c | 14 ++++++++++++++ > > 4 files changed, 52 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..fb3b614f6456 100644 > > --- a/kernel/fork.c > > +++ b/kernel/fork.c > > @@ -1685,8 +1685,41 @@ static void pidfd_show_fdinfo(struct seq_file *m, struct file *f) > > } > > #endif > > > > +static unsigned int pidfd_poll(struct file *file, struct poll_table_struct *pts) > > +{ > > + struct task_struct *task; > > + struct pid *pid; > > + int poll_flags = 0; > > + > > + /* > > + * tasklist_lock must be held because to avoid racing with > > + * changes in exit_state and wake up. Basically to avoid: > > + * > > + * P0: read exit_state = 0 > > + * P1: write exit_state = EXIT_DEAD > > + * P1: Do a wake up - wq is empty, so do nothing > > + * P0: Queue for polling - wait forever. > > + */ > > + read_lock(&tasklist_lock); > > + pid = file->private_data; > > + task = pid_task(pid, PIDTYPE_PID); > > + WARN_ON_ONCE(task && !thread_group_leader(task)); > > + > > + if (!task || (task->exit_state && thread_group_empty(task))) > > + poll_flags = POLLIN | POLLRDNORM; > > So we block until the thread-group is empty? Hm, the thread-group leader > remains in zombie state until all threads are gone. Should probably just > be a short comment somewhere that callers are only informed about a > whole thread-group exit and not about when the thread-group leader has > actually exited. Ok, I'll add a comment. > I would like the ability to extend this interface in the future to allow > for actually reading data from the pidfd on EPOLLIN. > POSIX specifies that POLLIN and POLLRDNORM are set even if the > message to be read is zero. So one cheap way of doing this would > probably be to do a 0 read/ioctl. That wouldn't hurt your very limited > usecase and people could test whether the read returned non-0 data and > if so they know this interface got extended. If we never extend it here > it won't matter. I am a bit confused. What specific changes to this patch are you proposing? This patch makes poll block until the process exits. In the future, we can make it unblock for a other reasons as well, that's fine with me. I don't see how this patch prevents such extensions. > > + if (!poll_flags) > > + poll_wait(file, &pid->wait_pidfd, pts); > > + > > + read_unlock(&tasklist_lock); > > + > > + 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..16e7718316e5 100644 > > --- a/kernel/signal.c > > +++ b/kernel/signal.c > > @@ -1800,6 +1800,17 @@ int send_sigqueue(struct sigqueue *q, struct pid *pid, enum pid_type type) > > return ret; > > } > > > > +static void do_notify_pidfd(struct task_struct *task) > > Maybe a short command that this helper can only be called when we know > that task is a thread-group leader wouldn't hurt so there's no confusion > later. Ok, will do. thanks, - Joel
On Thu, Apr 25, 2019 at 03:00:09PM -0400, Joel Fernandes (Google) wrote: > pidfd are file descriptors referring to a process created with the > CLONE_PIDFD clone(2) flag. Android low memory killer (LMK) needs pidfd > polling support to replace code that currently checks for existence of > /proc/pid for knowing that a process that is signalled to be killed has > died, which is both racy and slow. The pidfd poll approach is race-free, > and also allows the LMK to do other things (such as by polling on other > fds) while awaiting the process being killed to die. > > It prevents a situation where 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. > > In this patch, we follow the same existing mechanism in the kernel used > when the parent of the task group is to be notified (do_notify_parent). > This is when the tasks waiting on a poll of pidfd are also awakened. > > 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: luto@amacapital.net > Cc: rostedt@goodmis.org > Cc: dancol@google.com > Cc: sspatil@google.com > Cc: christian@brauner.io > Cc: jannh@google.com > Cc: surenb@google.com > Cc: timmurray@google.com > Cc: Jonathan Kowalski <bl0pbl33p@gmail.com> > Cc: torvalds@linux-foundation.org > Cc: kernel-team@android.com That should be of the form: Cc: First Name <email@address.com> > Co-developed-by: Daniel Colascione <dancol@google.com> Every CDB needs to come with a SOB. > Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org> > > --- > > 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 | 33 +++++++++++++++++++++++++++++++++ > kernel/pid.c | 2 ++ > kernel/signal.c | 14 ++++++++++++++ > 4 files changed, 52 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..fb3b614f6456 100644 > --- a/kernel/fork.c > +++ b/kernel/fork.c > @@ -1685,8 +1685,41 @@ static void pidfd_show_fdinfo(struct seq_file *m, struct file *f) > } > #endif > > +static unsigned int pidfd_poll(struct file *file, struct poll_table_struct *pts) > +{ > + struct task_struct *task; > + struct pid *pid; > + int poll_flags = 0; > + > + /* > + * tasklist_lock must be held because to avoid racing with > + * changes in exit_state and wake up. Basically to avoid: > + * > + * P0: read exit_state = 0 > + * P1: write exit_state = EXIT_DEAD > + * P1: Do a wake up - wq is empty, so do nothing > + * P0: Queue for polling - wait forever. > + */ > + read_lock(&tasklist_lock); > + pid = file->private_data; > + task = pid_task(pid, PIDTYPE_PID); > + WARN_ON_ONCE(task && !thread_group_leader(task)); > + > + if (!task || (task->exit_state && thread_group_empty(task))) > + poll_flags = POLLIN | POLLRDNORM; > + > + if (!poll_flags) > + poll_wait(file, &pid->wait_pidfd, pts); > + > + read_unlock(&tasklist_lock); > + > + 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..16e7718316e5 100644 > --- a/kernel/signal.c > +++ b/kernel/signal.c > @@ -1800,6 +1800,17 @@ 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; > + > + lockdep_assert_held(&tasklist_lock); > + > + pid = get_task_pid(task, PIDTYPE_PID); > + wake_up_all(&pid->wait_pidfd); > + put_pid(pid); > +} > + > /* > * Let a parent know about the death of a child. > * For a stopped/continued status change, use do_notify_parent_cldstop instead. > @@ -1823,6 +1834,9 @@ 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
On Fri, Apr 26, 2019 at 10:23:37AM -0400, Joel Fernandes wrote: > On Fri, Apr 26, 2019 at 12:24:04AM +0200, Christian Brauner wrote: > > On Thu, Apr 25, 2019 at 03:00:09PM -0400, Joel Fernandes (Google) wrote: > > > pidfd are file descriptors referring to a process created with the > > > CLONE_PIDFD clone(2) flag. Android low memory killer (LMK) needs pidfd > > > polling support to replace code that currently checks for existence of > > > /proc/pid for knowing that a process that is signalled to be killed has > > > died, which is both racy and slow. The pidfd poll approach is race-free, > > > and also allows the LMK to do other things (such as by polling on other > > > fds) while awaiting the process being killed to die. > > > > Thanks for the patch! > > > > Ok, let me be a little bit anal. > > Please start the commit message with what this patch does and then add > > The subject title is "Add polling support to pidfd", but ok I should write a > better commit message. Yeah, it's really just that we should really just have a simple paragraph that expresses this makes the codebase do X. > > > the justification why. You just say the "pidfd-poll" approach. You can > > probably assume that CLONE_PIDFD is available for this patch. So > > something like: > > > > "This patch makes pidfds pollable. Specifically, it allows listeners to > > be informed when the process the pidfd referes to exits. This patch only > > introduces the ability to poll thread-group leaders since pidfds > > currently can only reference those..." > > > > Then justify the use-case and then go into implementation details. > > That's usually how I would think about this: > > - Change the codebase to do X > > - Why do we need X > > - Are there any technical details worth mentioning in the commit message > > (- Are there any controversial points that people stumbled upon but that > > have been settled sufficiently.) > > Generally the "how" in the patch should be in the code, but ok. That's why I said: technical details that are worth mentioning. Sometimes you have controversial bits that are obviously to be understood in the code but it still might be worth explaining why one had to do it this way. Like say what we did for the pidfd_send_signal() thing where we explained why O_PATH is disallowed. > > I changed the first 3 paragraphs of the changelog to the following, is that > better? : > > 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. > > > > pidfd are file descriptors referring to a process created with the > > > CLONE_PIDFD clone(2) flag. Android low memory killer (LMK) needs pidfd > > > polling support to replace code that currently checks for existence of > > > /proc/pid for knowing that a process that is signalled to be killed has > > > died, which is both racy and slow. The pidfd poll approach is race-free, > > > and also allows the LMK to do other things (such as by polling on other > > > fds) while awaiting the process being killed to die. > > > > > > > > It prevents a situation where 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. > > > > > > In this patch, we follow the same existing mechanism in the kernel used > > > when the parent of the task group is to be notified (do_notify_parent). > > > This is when the tasks waiting on a poll of pidfd are also awakened. > > > > > > 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: luto@amacapital.net > > > Cc: rostedt@goodmis.org > > > Cc: dancol@google.com > > > Cc: sspatil@google.com > > > Cc: christian@brauner.io > > > Cc: jannh@google.com > > > Cc: surenb@google.com > > > Cc: timmurray@google.com > > > Cc: Jonathan Kowalski <bl0pbl33p@gmail.com> > > > Cc: torvalds@linux-foundation.org > > > Cc: kernel-team@android.com > > > > These should all be in the form: > > > > Cc: Firstname Lastname <email@address.com> > > If this bothers you too much, I can also just remove the CC list from the > changelog here, and include it in my invocation of git-send-email instead.. > but I have seen commits in the tree that don't follow this rule. Yeah, but they should. There are people with multiple emails over the years and they might not necessarily contain their first and last name. And I don't want to have to mailmap them or sm. Having their names in there just makes it easier. Also, every single other DCO-*related* line follows: Random J Developer <random@developer.example.org> This should too. If others are sloppy and allow this, fine. No reason we should. > > > > > There are people missing from the Cc that really should be there... > > If you look at the CC list of the email, people in the get_maintainer.pl > script were also added. I did run get_maintainer.pl and checkpatch. But ok, I > will add the folks you are suggesting as well. Thanks. get_maintainer.pl is not the last word. > > > Even though he usually doesn't respond that often, please Cc Al on this. > > If he responds it's usually rather important. > > No issues on that, but I am wondering if he should also be in MAINTAINERS > file somewhere such that get_maintainer.pl does pick him up. I added him. It's often not about someone being a maintainer but whether or not they have valuable input. "[...] This tag documents that potentially interested parties have been included in the discussion." > > > Oleg has reviewed your RFC patch quite substantially and given valuable > > feedback and has an opinion on this thing and is best acquainted with > > the exit code. So please add him to the Cc of the commit message in the > > appropriate form and also add him to the Cc of the thread. > > Done. Thanks! > > > Probably also want linux-api for good measure since a lot of people are > > subscribed that would care about pollable pidfds. I'd also add Kees > > since he had some interest in this work and David (Howells). > > Done, I added all of them and CC will go out to them next time. Thanks. Cool. That really wasn't a "you've done this wrong". It's rather really just to make sure that everyone who might catch a big f*ck up on our part has had a chance to tell us so. :) > > > > > > Co-developed-by: Daniel Colascione <dancol@google.com> > > > > Every CDB needs to give a SOB as well. > > Ok, done. thanks. Fwiw, I only learned this recently too. > > > > > > Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org> > > > > > > --- > > > > > > 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 | 33 +++++++++++++++++++++++++++++++++ > > > kernel/pid.c | 2 ++ > > > kernel/signal.c | 14 ++++++++++++++ > > > 4 files changed, 52 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..fb3b614f6456 100644 > > > --- a/kernel/fork.c > > > +++ b/kernel/fork.c > > > @@ -1685,8 +1685,41 @@ static void pidfd_show_fdinfo(struct seq_file *m, struct file *f) > > > } > > > #endif > > > > > > +static unsigned int pidfd_poll(struct file *file, struct poll_table_struct *pts) > > > +{ > > > + struct task_struct *task; > > > + struct pid *pid; > > > + int poll_flags = 0; > > > + > > > + /* > > > + * tasklist_lock must be held because to avoid racing with > > > + * changes in exit_state and wake up. Basically to avoid: > > > + * > > > + * P0: read exit_state = 0 > > > + * P1: write exit_state = EXIT_DEAD > > > + * P1: Do a wake up - wq is empty, so do nothing > > > + * P0: Queue for polling - wait forever. > > > + */ > > > + read_lock(&tasklist_lock); > > > + pid = file->private_data; > > > + task = pid_task(pid, PIDTYPE_PID); > > > + WARN_ON_ONCE(task && !thread_group_leader(task)); > > > + > > > + if (!task || (task->exit_state && thread_group_empty(task))) > > > + poll_flags = POLLIN | POLLRDNORM; > > > > So we block until the thread-group is empty? Hm, the thread-group leader > > remains in zombie state until all threads are gone. Should probably just > > be a short comment somewhere that callers are only informed about a > > whole thread-group exit and not about when the thread-group leader has > > actually exited. > > Ok, I'll add a comment. > > > I would like the ability to extend this interface in the future to allow > > for actually reading data from the pidfd on EPOLLIN. > > POSIX specifies that POLLIN and POLLRDNORM are set even if the > > message to be read is zero. So one cheap way of doing this would > > probably be to do a 0 read/ioctl. That wouldn't hurt your very limited > > usecase and people could test whether the read returned non-0 data and > > if so they know this interface got extended. If we never extend it here > > it won't matter. > > I am a bit confused. What specific changes to this patch are you proposing? > This patch makes poll block until the process exits. In the future, we can > make it unblock for a other reasons as well, that's fine with me. I don't see > how this patch prevents such extensions. I guess I should've asked the following: What happens right now, when you get EPOLLIN on the pidfd and you and out of ignorance you do: read(pidfd, ...) > > > > + if (!poll_flags) > > > + poll_wait(file, &pid->wait_pidfd, pts); > > > + > > > + read_unlock(&tasklist_lock); > > > + > > > + 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..16e7718316e5 100644 > > > --- a/kernel/signal.c > > > +++ b/kernel/signal.c > > > @@ -1800,6 +1800,17 @@ int send_sigqueue(struct sigqueue *q, struct pid *pid, enum pid_type type) > > > return ret; > > > } > > > > > > +static void do_notify_pidfd(struct task_struct *task) > > > > Maybe a short command that this helper can only be called when we know > > that task is a thread-group leader wouldn't hurt so there's no confusion > > later. > > Ok, will do. > > thanks, > > - Joel >
On April 26, 2019 5:21:40 PM GMT+02:00, Christian Brauner <christian@brauner.io> wrote: >On Fri, Apr 26, 2019 at 10:23:37AM -0400, Joel Fernandes wrote: >> On Fri, Apr 26, 2019 at 12:24:04AM +0200, Christian Brauner wrote: >> > On Thu, Apr 25, 2019 at 03:00:09PM -0400, Joel Fernandes (Google) >wrote: >> > > pidfd are file descriptors referring to a process created with >the >> > > CLONE_PIDFD clone(2) flag. Android low memory killer (LMK) needs >pidfd >> > > polling support to replace code that currently checks for >existence of >> > > /proc/pid for knowing that a process that is signalled to be >killed has >> > > died, which is both racy and slow. The pidfd poll approach is >race-free, >> > > and also allows the LMK to do other things (such as by polling on >other >> > > fds) while awaiting the process being killed to die. >> > >> > Thanks for the patch! >> > >> > Ok, let me be a little bit anal. >> > Please start the commit message with what this patch does and then >add >> >> The subject title is "Add polling support to pidfd", but ok I should >write a >> better commit message. > >Yeah, it's really just that we should really just have a simple >paragraph that expresses this makes the codebase do X. > >> >> > the justification why. You just say the "pidfd-poll" approach. You >can >> > probably assume that CLONE_PIDFD is available for this patch. So >> > something like: >> > >> > "This patch makes pidfds pollable. Specifically, it allows >listeners to >> > be informed when the process the pidfd referes to exits. This patch >only >> > introduces the ability to poll thread-group leaders since pidfds >> > currently can only reference those..." >> > >> > Then justify the use-case and then go into implementation details. >> > That's usually how I would think about this: >> > - Change the codebase to do X >> > - Why do we need X >> > - Are there any technical details worth mentioning in the commit >message >> > (- Are there any controversial points that people stumbled upon but >that >> > have been settled sufficiently.) >> >> Generally the "how" in the patch should be in the code, but ok. > >That's why I said: technical details that are worth mentioning. >Sometimes you have controversial bits that are obviously to be >understood in the code but it still might be worth explaining why one >had to do it this way. Like say what we did for the pidfd_send_signal() >thing where we explained why O_PATH is disallowed. > >> >> I changed the first 3 paragraphs of the changelog to the following, >is that >> better? : >> >> 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. >> >> > > pidfd are file descriptors referring to a process created with >the >> > > CLONE_PIDFD clone(2) flag. Android low memory killer (LMK) needs >pidfd >> > > polling support to replace code that currently checks for >existence of >> > > /proc/pid for knowing that a process that is signalled to be >killed has >> > > died, which is both racy and slow. The pidfd poll approach is >race-free, >> > > and also allows the LMK to do other things (such as by polling on >other >> > > fds) while awaiting the process being killed to die. >> > >> > > >> > > It prevents a situation where 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. >> > > >> > > In this patch, we follow the same existing mechanism in the >kernel used >> > > when the parent of the task group is to be notified >(do_notify_parent). >> > > This is when the tasks waiting on a poll of pidfd are also >awakened. >> > > >> > > 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: luto@amacapital.net >> > > Cc: rostedt@goodmis.org >> > > Cc: dancol@google.com >> > > Cc: sspatil@google.com >> > > Cc: christian@brauner.io >> > > Cc: jannh@google.com >> > > Cc: surenb@google.com >> > > Cc: timmurray@google.com >> > > Cc: Jonathan Kowalski <bl0pbl33p@gmail.com> >> > > Cc: torvalds@linux-foundation.org >> > > Cc: kernel-team@android.com >> > >> > These should all be in the form: >> > >> > Cc: Firstname Lastname <email@address.com> >> >> If this bothers you too much, I can also just remove the CC list from >the >> changelog here, and include it in my invocation of git-send-email >instead.. >> but I have seen commits in the tree that don't follow this rule. > >Yeah, but they should. There are people with multiple emails over the >years and they might not necessarily contain their first and last >name. And I don't want to have to mailmap them or sm. Having their >names >in there just makes it easier. Also, every single other DCO-*related* >line follows: > >Random J Developer <random@developer.example.org> > >This should too. If others are sloppy and allow this, fine. No reason >we >should. > >> >> > >> > There are people missing from the Cc that really should be there... >> >> If you look at the CC list of the email, people in the >get_maintainer.pl >> script were also added. I did run get_maintainer.pl and checkpatch. >But ok, I >> will add the folks you are suggesting as well. Thanks. > >get_maintainer.pl is not the last word. > >> >> > Even though he usually doesn't respond that often, please Cc Al on >this. >> > If he responds it's usually rather important. >> >> No issues on that, but I am wondering if he should also be in >MAINTAINERS >> file somewhere such that get_maintainer.pl does pick him up. I added >him. > >It's often not about someone being a maintainer but whether or not they >have valuable input. > >"[...] This tag documents that potentially interested parties have been >included in the discussion." > >> >> > Oleg has reviewed your RFC patch quite substantially and given >valuable >> > feedback and has an opinion on this thing and is best acquainted >with >> > the exit code. So please add him to the Cc of the commit message in >the >> > appropriate form and also add him to the Cc of the thread. >> >> Done. > >Thanks! > >> >> > Probably also want linux-api for good measure since a lot of people >are >> > subscribed that would care about pollable pidfds. I'd also add Kees >> > since he had some interest in this work and David (Howells). >> >> Done, I added all of them and CC will go out to them next time. >Thanks. > >Cool. That really wasn't a "you've done this wrong". It's rather really >just to make sure that everyone who might catch a big f*ck up on our >part has had a chance to tell us so. :) > >> >> > >> > > Co-developed-by: Daniel Colascione <dancol@google.com> >> > >> > Every CDB needs to give a SOB as well. >> >> Ok, done. thanks. > >Fwiw, I only learned this recently too. > >> >> > >> > > Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org> >> > > >> > > --- >> > > >> > > 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 | 33 +++++++++++++++++++++++++++++++++ >> > > kernel/pid.c | 2 ++ >> > > kernel/signal.c | 14 ++++++++++++++ >> > > 4 files changed, 52 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..fb3b614f6456 100644 >> > > --- a/kernel/fork.c >> > > +++ b/kernel/fork.c >> > > @@ -1685,8 +1685,41 @@ static void pidfd_show_fdinfo(struct >seq_file *m, struct file *f) >> > > } >> > > #endif >> > > >> > > +static unsigned int pidfd_poll(struct file *file, struct >poll_table_struct *pts) >> > > +{ >> > > + struct task_struct *task; >> > > + struct pid *pid; >> > > + int poll_flags = 0; >> > > + >> > > + /* >> > > + * tasklist_lock must be held because to avoid racing with >> > > + * changes in exit_state and wake up. Basically to avoid: >> > > + * >> > > + * P0: read exit_state = 0 >> > > + * P1: write exit_state = EXIT_DEAD >> > > + * P1: Do a wake up - wq is empty, so do nothing >> > > + * P0: Queue for polling - wait forever. >> > > + */ >> > > + read_lock(&tasklist_lock); >> > > + pid = file->private_data; >> > > + task = pid_task(pid, PIDTYPE_PID); >> > > + WARN_ON_ONCE(task && !thread_group_leader(task)); >> > > + >> > > + if (!task || (task->exit_state && thread_group_empty(task))) >> > > + poll_flags = POLLIN | POLLRDNORM; >> > >> > So we block until the thread-group is empty? Hm, the thread-group >leader >> > remains in zombie state until all threads are gone. Should probably >just >> > be a short comment somewhere that callers are only informed about a >> > whole thread-group exit and not about when the thread-group leader >has >> > actually exited. >> >> Ok, I'll add a comment. >> >> > I would like the ability to extend this interface in the future to >allow >> > for actually reading data from the pidfd on EPOLLIN. >> > POSIX specifies that POLLIN and POLLRDNORM are set even if the >> > message to be read is zero. So one cheap way of doing this would >> > probably be to do a 0 read/ioctl. That wouldn't hurt your very >limited >> > usecase and people could test whether the read returned non-0 data >and >> > if so they know this interface got extended. If we never extend it >here >> > it won't matter. >> >> I am a bit confused. What specific changes to this patch are you >proposing? >> This patch makes poll block until the process exits. In the future, >we can >> make it unblock for a other reasons as well, that's fine with me. I >don't see >> how this patch prevents such extensions. > >I guess I should've asked the following: >What happens right now, when you get EPOLLIN on the pidfd and you and >out of ignorance you do: > >read(pidfd, ...) I guess it returns EINVAL which is fine. So you can ignore that comment. > >> >> > > + if (!poll_flags) >> > > + poll_wait(file, &pid->wait_pidfd, pts); >> > > + >> > > + read_unlock(&tasklist_lock); >> > > + >> > > + 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..16e7718316e5 100644 >> > > --- a/kernel/signal.c >> > > +++ b/kernel/signal.c >> > > @@ -1800,6 +1800,17 @@ int send_sigqueue(struct sigqueue *q, >struct pid *pid, enum pid_type type) >> > > return ret; >> > > } >> > > >> > > +static void do_notify_pidfd(struct task_struct *task) >> > >> > Maybe a short command that this helper can only be called when we >know >> > that task is a thread-group leader wouldn't hurt so there's no >confusion >> > later. >> >> Ok, will do. >> >> thanks, >> >> - Joel >>
Thanks for cc'ing me... On 04/26, Christian Brauner wrote: > > On Thu, Apr 25, 2019 at 03:00:09PM -0400, Joel Fernandes (Google) wrote: > > +static unsigned int pidfd_poll(struct file *file, struct poll_table_struct *pts) > > +{ > > + struct task_struct *task; > > + struct pid *pid; > > + int poll_flags = 0; > > + > > + /* > > + * tasklist_lock must be held because to avoid racing with > > + * changes in exit_state and wake up. Basically to avoid: > > + * > > + * P0: read exit_state = 0 > > + * P1: write exit_state = EXIT_DEAD > > + * P1: Do a wake up - wq is empty, so do nothing > > + * P0: Queue for polling - wait forever. > > + */ > > + read_lock(&tasklist_lock); > > + pid = file->private_data; > > + task = pid_task(pid, PIDTYPE_PID); > > + WARN_ON_ONCE(task && !thread_group_leader(task)); > > + > > + if (!task || (task->exit_state && thread_group_empty(task))) > > + poll_flags = POLLIN | POLLRDNORM; Joel, I still can't understand why do we need tasklist... and I don't really understand the comment. The code looks as if you are trying to avoid poll_wait(), but this would be strange. OK, why can't pidfd_poll() do poll_wait(file, &pid->wait_pidfd, pts); rcu_read_lock(); task = pid_task(pid, PIDTYPE_PID); if (!task || task->exit_state && thread_group_empty(task)) poll_flags = POLLIN | ...; rcu_read_unlock(); return poll_flags; ? > > +static void do_notify_pidfd(struct task_struct *task) > > Maybe a short command that this helper can only be called when we know > that task is a thread-group leader wouldn't hurt so there's no confusion > later. Not really. If the task is traced, do_notify_parent() (and thus do_notify_pidfd()) can be called to notify the debugger even if the task is not a leader and/or if it is not the last thread. The latter means a spurious wakeup for pidfd_poll(). > > +{ > > + struct pid *pid; > > + > > + lockdep_assert_held(&tasklist_lock); > > + > > + pid = get_task_pid(task, PIDTYPE_PID); > > + wake_up_all(&pid->wait_pidfd); > > + put_pid(pid); Why get/put? Oleg.
On Sun, Apr 28, 2019 at 06:24:06PM +0200, Oleg Nesterov wrote: > Thanks for cc'ing me... > > On 04/26, Christian Brauner wrote: > > > > On Thu, Apr 25, 2019 at 03:00:09PM -0400, Joel Fernandes (Google) wrote: > > > +static unsigned int pidfd_poll(struct file *file, struct poll_table_struct *pts) > > > +{ > > > + struct task_struct *task; > > > + struct pid *pid; > > > + int poll_flags = 0; > > > + > > > + /* > > > + * tasklist_lock must be held because to avoid racing with > > > + * changes in exit_state and wake up. Basically to avoid: > > > + * > > > + * P0: read exit_state = 0 > > > + * P1: write exit_state = EXIT_DEAD > > > + * P1: Do a wake up - wq is empty, so do nothing > > > + * P0: Queue for polling - wait forever. > > > + */ > > > + read_lock(&tasklist_lock); > > > + pid = file->private_data; > > > + task = pid_task(pid, PIDTYPE_PID); > > > + WARN_ON_ONCE(task && !thread_group_leader(task)); > > > + > > > + if (!task || (task->exit_state && thread_group_empty(task))) > > > + poll_flags = POLLIN | POLLRDNORM; > > Joel, I still can't understand why do we need tasklist... and I don't really > understand the comment. The code looks as if you are trying to avoid poll_wait(), > but this would be strange. > > OK, why can't pidfd_poll() do > > poll_wait(file, &pid->wait_pidfd, pts); > > rcu_read_lock(); > task = pid_task(pid, PIDTYPE_PID); > if (!task || task->exit_state && thread_group_empty(task)) > poll_flags = POLLIN | ...; > rcu_read_unlock(); > > return poll_flags; > > ? Oh that's much better Oleg, and would avoid the race I had in mind: Basically I was acquiring the tasklist_lock to avoid a case where a polling task is not woken up because it was added to the waitqueue too late. The reading of the exit_state and the conditional adding to the wait queue, needed to be atomic. Otherwise something like the following may be possible: Task A (poller) Task B (exiting task being polled) ------------ ---------------- poll() called exit_state is set to non-zero read exit_state wake_up_all() add_wait_queue() ---------------------------------------------- However, in your code above, it is avoided because we get: Task A (poller) Task B (exiting task being polled) ------------ ---------------- poll() called add_wait_queue() exit_state is set to non-zero read exit_state remove_wait_queue() wake_up_all() I don't see any other issues with your code above so I can try it out and update the patches. Thanks. > > > +static void do_notify_pidfd(struct task_struct *task) > > > > Maybe a short command that this helper can only be called when we know > > that task is a thread-group leader wouldn't hurt so there's no confusion > > later. > > Not really. If the task is traced, do_notify_parent() (and thus do_notify_pidfd()) > can be called to notify the debugger even if the task is not a leader and/or if > it is not the last thread. The latter means a spurious wakeup for pidfd_poll(). Seems like you are replying to Christian's point. I agree with you. > > > +{ > > > + struct pid *pid; > > > + > > > + lockdep_assert_held(&tasklist_lock); > > > + > > > + pid = get_task_pid(task, PIDTYPE_PID); > > > + wake_up_all(&pid->wait_pidfd); > > > + put_pid(pid); > > Why get/put? Yes, pid_task() should do it. Will update it. Thanks! - Joel
On Mon, Apr 29, 2019 at 10:02:45AM -0400, Joel Fernandes wrote: > On Sun, Apr 28, 2019 at 06:24:06PM +0200, Oleg Nesterov wrote: [snip] > > > > +{ > > > > + struct pid *pid; > > > > + > > > > + lockdep_assert_held(&tasklist_lock); > > > > + > > > > + pid = get_task_pid(task, PIDTYPE_PID); > > > > + wake_up_all(&pid->wait_pidfd); > > > > + put_pid(pid); > > > > Why get/put? > > Yes, pid_task() should do it. Will update it. Thanks! I spoke too soon. We need the task's pid of type PIDTYPE_PID. How else can we get it? This does an atomic_inc on the pid->count, so we need to put_pid() after we are done with it. Did I miss something? thanks, - Joel
On 04/29, Joel Fernandes wrote: > > However, in your code above, it is avoided because we get: > > Task A (poller) Task B (exiting task being polled) > ------------ ---------------- > poll() called > add_wait_queue() > exit_state is set to non-zero > read exit_state > remove_wait_queue() > wake_up_all() just to clarify... No, sys_poll() path doesn't do remove_wait_queue() until it returns to user mode, and that is why we can't race with set-exit_code + wake_up(). pidfd_poll() can race with the exiting task, miss exit_code != 0, and return zero. However, do_poll() won't block after that and pidfd_poll() will be called again. Oleg.
On 04/29, Joel Fernandes wrote: > > On Mon, Apr 29, 2019 at 10:02:45AM -0400, Joel Fernandes wrote: > > On Sun, Apr 28, 2019 at 06:24:06PM +0200, Oleg Nesterov wrote: > [snip] > > > > > +{ > > > > > + struct pid *pid; > > > > > + > > > > > + lockdep_assert_held(&tasklist_lock); > > > > > + > > > > > + pid = get_task_pid(task, PIDTYPE_PID); > > > > > + wake_up_all(&pid->wait_pidfd); > > > > > + put_pid(pid); > > > > > > Why get/put? > > > > Yes, pid_task() should do it. Will update it. Thanks! > > I spoke too soon. We need the task's pid of type PIDTYPE_PID. How else can we > get it? This does an atomic_inc on the pid->count, so we need to put_pid() > after we are done with it. Did I miss something? Just use task_pid(task); Oleg.
On Mon, Apr 29, 2019 at 04:20:30PM +0200, Oleg Nesterov wrote: > On 04/29, Joel Fernandes wrote: > > > > However, in your code above, it is avoided because we get: > > > > Task A (poller) Task B (exiting task being polled) > > ------------ ---------------- > > poll() called > > add_wait_queue() > > exit_state is set to non-zero > > read exit_state > > remove_wait_queue() > > wake_up_all() > > just to clarify... No, sys_poll() path doesn't do remove_wait_queue() until > it returns to user mode, and that is why we can't race with set-exit_code + > wake_up(). I didn't follow what you mean, the removal from the waitqueue happens in free_poll_entry() called from poll_freewait() which happens from do_sys_poll() which is before the syscall returns to user mode. Could you explain more? > pidfd_poll() can race with the exiting task, miss exit_code != 0, and return > zero. However, do_poll() won't block after that and pidfd_poll() will be called > again. Here also I didn't follow what you mean. If exit_code is read as 0 in pidfd_poll(), then in do_poll() the count will be 0 and it will block in poll_schedule_timeout(). Right? But above you're saying it wont block. Also if you could show a timing diagram of this different race you're talking about, that will make things clear. It is a bit hard for me to picture otherwise. Also, I will use task_pid() for getting the pid from the task, as you suggest in the other thread. thanks, - Joel
On 04/29, Joel Fernandes wrote: > > On Mon, Apr 29, 2019 at 04:20:30PM +0200, Oleg Nesterov wrote: > > On 04/29, Joel Fernandes wrote: > > > > > > However, in your code above, it is avoided because we get: > > > > > > Task A (poller) Task B (exiting task being polled) > > > ------------ ---------------- > > > poll() called > > > add_wait_queue() > > > exit_state is set to non-zero > > > read exit_state > > > remove_wait_queue() > > > wake_up_all() > > > > just to clarify... No, sys_poll() path doesn't do remove_wait_queue() until > > it returns to user mode, and that is why we can't race with set-exit_code + > > wake_up(). > > I didn't follow what you mean, the removal from the waitqueue happens in > free_poll_entry() called from poll_freewait() which happens from > do_sys_poll() which is before the syscall returns to user mode. Could you > explain more? Hmm. I do not really understand the question... Sure, do_sys_poll() does poll_freewait() before sysret or even before return from syscall, but why does this matter? This is the exit path, it frees the memory, does fput(), etc, f_op->poll() won't be call after that. > > pidfd_poll() can race with the exiting task, miss exit_code != 0, and return > > zero. However, do_poll() won't block after that and pidfd_poll() will be called > > again. > > Here also I didn't follow what you mean. If exit_code is read as 0 in > pidfd_poll(), then in do_poll() the count will be 0 and it will block in > poll_schedule_timeout(). Right? No. Please note the pwq->triggered check and please read __pollwake(). But if you want to understand this you can forget about poll/select. It is a bit complicated, in particular because it has to do set_current_state() right before schedule() and thus it plays games with pwq->triggered. But in essence this doesn't differ too much from the plain wait_event-like code (although you can also look at wait_woken/woken_wake_function). If remove_wait_queue() could happem before wake_up_all() (like in your pseudo- code above), then pidfd_poll() or any other ->poll() method could miss _both_ the condition and wakeup. But sys_poll() doesn't do this, so it is fine to miss the condition and rely on wake_up_all() which ensures we won't block and the next iteration must see condition == T. Oleg.
On 04/30, Oleg Nesterov wrote: > > > > pidfd_poll() can race with the exiting task, miss exit_code != 0, and return > > > zero. However, do_poll() won't block after that and pidfd_poll() will be called > > > again. > > > > Here also I didn't follow what you mean. If exit_code is read as 0 in > > pidfd_poll(), then in do_poll() the count will be 0 and it will block in > > poll_schedule_timeout(). Right? > > No. Please note the pwq->triggered check and please read __pollwake(). > > But if you want to understand this you can forget about poll/select. It is > a bit complicated, in particular because it has to do set_current_state() > right before schedule() and thus it plays games with pwq->triggered. But in > essence this doesn't differ too much from the plain wait_event-like code > (although you can also look at wait_woken/woken_wake_function). > > If remove_wait_queue() could happem before wake_up_all() (like in your pseudo- > code above), then pidfd_poll() or any other ->poll() method could miss _both_ > the condition and wakeup. But sys_poll() doesn't do this, so it is fine to miss > the condition and rely on wake_up_all() which ensures we won't block and the > next iteration must see condition == T. Oh, just in case... If it is not clear, of course I am talking about the case when wake_up_call() was already called when we check the condition. Otherwise everything is simple. Oleg.
On Tue, Apr 30, 2019 at 01:53:33PM +0200, Oleg Nesterov wrote: > On 04/29, Joel Fernandes wrote: > > > > On Mon, Apr 29, 2019 at 04:20:30PM +0200, Oleg Nesterov wrote: > > > On 04/29, Joel Fernandes wrote: > > > > > > > > However, in your code above, it is avoided because we get: > > > > > > > > Task A (poller) Task B (exiting task being polled) > > > > ------------ ---------------- > > > > poll() called > > > > add_wait_queue() > > > > exit_state is set to non-zero > > > > read exit_state > > > > remove_wait_queue() > > > > wake_up_all() > > > > > > just to clarify... No, sys_poll() path doesn't do remove_wait_queue() until > > > it returns to user mode, and that is why we can't race with set-exit_code + > > > wake_up(). > > > > I didn't follow what you mean, the removal from the waitqueue happens in > > free_poll_entry() called from poll_freewait() which happens from > > do_sys_poll() which is before the syscall returns to user mode. Could you > > explain more? > > Hmm. I do not really understand the question... Sure, do_sys_poll() does > poll_freewait() before sysret or even before return from syscall, but why > does this matter? This is the exit path, it frees the memory, does fput(), > etc, f_op->poll() won't be call after that. Ok, we are on the same page on this. > > > pidfd_poll() can race with the exiting task, miss exit_code != 0, and return > > > zero. However, do_poll() won't block after that and pidfd_poll() will be called > > > again. > > > > Here also I didn't follow what you mean. If exit_code is read as 0 in > > pidfd_poll(), then in do_poll() the count will be 0 and it will block in > > poll_schedule_timeout(). Right? > > No. Please note the pwq->triggered check and please read __pollwake(). > > But if you want to understand this you can forget about poll/select. It is > a bit complicated, in particular because it has to do set_current_state() > right before schedule() and thus it plays games with pwq->triggered. But in > essence this doesn't differ too much from the plain wait_event-like code > (although you can also look at wait_woken/woken_wake_function). > > If remove_wait_queue() could happem before wake_up_all() (like in your pseudo- > code above), then pidfd_poll() or any other ->poll() method could miss _both_ > the condition and wakeup. But sys_poll() doesn't do this, so it is fine to miss > the condition and rely on wake_up_all() which ensures we won't block and the > next iteration must see condition == T. Agreed. In my pseudo-code above, I meant removal from waitqueue only once we are not going to be blocking in poll and returning to userspace. I may have messed the sequence of events, but my point was to show the race I had in mind (missing a wake up due to adding to the waitqueue too late inside pidfd_poll()). Anyway, I will repost with your suggested change and send it soon. Thanks for the discussions. thanks, - Joel
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..fb3b614f6456 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -1685,8 +1685,41 @@ static void pidfd_show_fdinfo(struct seq_file *m, struct file *f) } #endif +static unsigned int pidfd_poll(struct file *file, struct poll_table_struct *pts) +{ + struct task_struct *task; + struct pid *pid; + int poll_flags = 0; + + /* + * tasklist_lock must be held because to avoid racing with + * changes in exit_state and wake up. Basically to avoid: + * + * P0: read exit_state = 0 + * P1: write exit_state = EXIT_DEAD + * P1: Do a wake up - wq is empty, so do nothing + * P0: Queue for polling - wait forever. + */ + read_lock(&tasklist_lock); + pid = file->private_data; + task = pid_task(pid, PIDTYPE_PID); + WARN_ON_ONCE(task && !thread_group_leader(task)); + + if (!task || (task->exit_state && thread_group_empty(task))) + poll_flags = POLLIN | POLLRDNORM; + + if (!poll_flags) + poll_wait(file, &pid->wait_pidfd, pts); + + read_unlock(&tasklist_lock); + + 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..16e7718316e5 100644 --- a/kernel/signal.c +++ b/kernel/signal.c @@ -1800,6 +1800,17 @@ 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; + + lockdep_assert_held(&tasklist_lock); + + pid = get_task_pid(task, PIDTYPE_PID); + wake_up_all(&pid->wait_pidfd); + put_pid(pid); +} + /* * Let a parent know about the death of a child. * For a stopped/continued status change, use do_notify_parent_cldstop instead. @@ -1823,6 +1834,9 @@ 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.
pidfd are file descriptors referring to a process created with the CLONE_PIDFD clone(2) flag. Android low memory killer (LMK) needs pidfd polling support to replace code that currently checks for existence of /proc/pid for knowing that a process that is signalled to be killed has died, which is both racy and slow. The pidfd poll approach is race-free, and also allows the LMK to do other things (such as by polling on other fds) while awaiting the process being killed to die. It prevents a situation where 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. In this patch, we follow the same existing mechanism in the kernel used when the parent of the task group is to be notified (do_notify_parent). This is when the tasks waiting on a poll of pidfd are also awakened. 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: luto@amacapital.net Cc: rostedt@goodmis.org Cc: dancol@google.com Cc: sspatil@google.com Cc: christian@brauner.io Cc: jannh@google.com Cc: surenb@google.com Cc: timmurray@google.com Cc: Jonathan Kowalski <bl0pbl33p@gmail.com> Cc: torvalds@linux-foundation.org Cc: kernel-team@android.com Co-developed-by: Daniel Colascione <dancol@google.com> Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org> --- 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 | 33 +++++++++++++++++++++++++++++++++ kernel/pid.c | 2 ++ kernel/signal.c | 14 ++++++++++++++ 4 files changed, 52 insertions(+)