| Message ID | 20210804185004.1304692-1-surenb@google.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | [v6,1/2] mm: introduce process_mrelease system call | expand |
On 04.08.21 20:50, Suren Baghdasaryan wrote: > In modern systems it's not unusual to have a system component monitoring > memory conditions of the system and tasked with keeping system memory > pressure under control. One way to accomplish that is to kill > non-essential processes to free up memory for more important ones. > Examples of this are Facebook's OOM killer daemon called oomd and > Android's low memory killer daemon called lmkd. > For such system component it's important to be able to free memory > quickly and efficiently. Unfortunately the time process takes to free > up its memory after receiving a SIGKILL might vary based on the state > of the process (uninterruptible sleep), size and OPP level of the core > the process is running. A mechanism to free resources of the target > process in a more predictable way would improve system's ability to > control its memory pressure. > Introduce process_mrelease system call that releases memory of a dying > process from the context of the caller. This way the memory is freed in > a more controllable way with CPU affinity and priority of the caller. > The workload of freeing the memory will also be charged to the caller. > The operation is allowed only on a dying process. > > After previous discussions [1, 2, 3] the decision was made [4] to introduce > a dedicated system call to cover this use case. > > The API is as follows, > > int process_mrelease(int pidfd, unsigned int flags); > > DESCRIPTION > The process_mrelease() system call is used to free the memory of > an exiting process. > > The pidfd selects the process referred to by the PID file > descriptor. > (See pidofd_open(2) for further information) > > The flags argument is reserved for future use; currently, this > argument must be specified as 0. > > RETURN VALUE > On success, process_mrelease() returns 0. On error, -1 is > returned and errno is set to indicate the error. > > ERRORS > EBADF pidfd is not a valid PID file descriptor. > > EAGAIN Failed to release part of the address space. > > EINTR The call was interrupted by a signal; see signal(7). > > EINVAL flags is not 0. > > EINVAL The memory of the task cannot be released because the > process is not exiting, the address space is shared > with another live process or there is a core dump in > progress. > > ENOSYS This system call is not supported, for example, without > MMU support built into Linux. > > ESRCH The target process does not exist (i.e., it has terminated > and been waited on). > > [1] https://lore.kernel.org/lkml/20190411014353.113252-3-surenb@google.com/ > [2] https://lore.kernel.org/linux-api/20201113173448.1863419-1-surenb@google.com/ > [3] https://lore.kernel.org/linux-api/20201124053943.1684874-3-surenb@google.com/ > [4] https://lore.kernel.org/linux-api/20201223075712.GA4719@lst.de/ > Acked-by: David Hildenbrand <david@redhat.com>
On Wed, 4 Aug 2021 11:50:03 -0700 Suren Baghdasaryan <surenb@google.com> wrote: > In modern systems it's not unusual to have a system component monitoring > memory conditions of the system and tasked with keeping system memory > pressure under control. One way to accomplish that is to kill > non-essential processes to free up memory for more important ones. > Examples of this are Facebook's OOM killer daemon called oomd and > Android's low memory killer daemon called lmkd. > For such system component it's important to be able to free memory > quickly and efficiently. Unfortunately the time process takes to free > up its memory after receiving a SIGKILL might vary based on the state > of the process (uninterruptible sleep), size and OPP level of the core > the process is running. A mechanism to free resources of the target > process in a more predictable way would improve system's ability to > control its memory pressure. > Introduce process_mrelease system call that releases memory of a dying > process from the context of the caller. This way the memory is freed in > a more controllable way with CPU affinity and priority of the caller. > The workload of freeing the memory will also be charged to the caller. > The operation is allowed only on a dying process. > > After previous discussions [1, 2, 3] the decision was made [4] to introduce > a dedicated system call to cover this use case. > > The API is as follows, > > int process_mrelease(int pidfd, unsigned int flags); > > DESCRIPTION > The process_mrelease() system call is used to free the memory of > an exiting process. > > The pidfd selects the process referred to by the PID file > descriptor. > (See pidofd_open(2) for further information) I did s/pidofd_open/pidfd_open/ > > The flags argument is reserved for future use; currently, this > argument must be specified as 0. > > RETURN VALUE > On success, process_mrelease() returns 0. On error, -1 is > returned and errno is set to indicate the error. > > ERRORS > EBADF pidfd is not a valid PID file descriptor. > > EAGAIN Failed to release part of the address space. > > EINTR The call was interrupted by a signal; see signal(7). > > EINVAL flags is not 0. > > EINVAL The memory of the task cannot be released because the > process is not exiting, the address space is shared > with another live process or there is a core dump in > progress. > > ENOSYS This system call is not supported, for example, without > MMU support built into Linux. > > ESRCH The target process does not exist (i.e., it has terminated > and been waited on). > > ... > > mm/oom_kill.c | 65 +++++++++++++++++++++++++++++++++++++++++++++++++++ > 1 file changed, 65 insertions(+) The code is nice and simple. Can we get a test suite into tools/testing/selftests?
On Wed, Aug 4, 2021 at 3:50 PM Andrew Morton <akpm@linux-foundation.org> wrote: > > On Wed, 4 Aug 2021 11:50:03 -0700 Suren Baghdasaryan <surenb@google.com> wrote: > > > In modern systems it's not unusual to have a system component monitoring > > memory conditions of the system and tasked with keeping system memory > > pressure under control. One way to accomplish that is to kill > > non-essential processes to free up memory for more important ones. > > Examples of this are Facebook's OOM killer daemon called oomd and > > Android's low memory killer daemon called lmkd. > > For such system component it's important to be able to free memory > > quickly and efficiently. Unfortunately the time process takes to free > > up its memory after receiving a SIGKILL might vary based on the state > > of the process (uninterruptible sleep), size and OPP level of the core > > the process is running. A mechanism to free resources of the target > > process in a more predictable way would improve system's ability to > > control its memory pressure. > > Introduce process_mrelease system call that releases memory of a dying > > process from the context of the caller. This way the memory is freed in > > a more controllable way with CPU affinity and priority of the caller. > > The workload of freeing the memory will also be charged to the caller. > > The operation is allowed only on a dying process. > > > > After previous discussions [1, 2, 3] the decision was made [4] to introduce > > a dedicated system call to cover this use case. > > > > The API is as follows, > > > > int process_mrelease(int pidfd, unsigned int flags); > > > > DESCRIPTION > > The process_mrelease() system call is used to free the memory of > > an exiting process. > > > > The pidfd selects the process referred to by the PID file > > descriptor. > > (See pidofd_open(2) for further information) > > I did s/pidofd_open/pidfd_open/ Thanks! > > > > > The flags argument is reserved for future use; currently, this > > argument must be specified as 0. > > > > RETURN VALUE > > On success, process_mrelease() returns 0. On error, -1 is > > returned and errno is set to indicate the error. > > > > ERRORS > > EBADF pidfd is not a valid PID file descriptor. > > > > EAGAIN Failed to release part of the address space. > > > > EINTR The call was interrupted by a signal; see signal(7). > > > > EINVAL flags is not 0. > > > > EINVAL The memory of the task cannot be released because the > > process is not exiting, the address space is shared > > with another live process or there is a core dump in > > progress. > > > > ENOSYS This system call is not supported, for example, without > > MMU support built into Linux. > > > > ESRCH The target process does not exist (i.e., it has terminated > > and been waited on). > > > > ... > > > > mm/oom_kill.c | 65 +++++++++++++++++++++++++++++++++++++++++++++++++++ > > 1 file changed, 65 insertions(+) > > The code is nice and simple. > > Can we get a test suite into tools/testing/selftests? Let me take a stab at it. Thanks!
On Wed, Aug 4, 2021 at 11:50 AM Suren Baghdasaryan <surenb@google.com> wrote: > > In modern systems it's not unusual to have a system component monitoring > memory conditions of the system and tasked with keeping system memory > pressure under control. One way to accomplish that is to kill > non-essential processes to free up memory for more important ones. > Examples of this are Facebook's OOM killer daemon called oomd and > Android's low memory killer daemon called lmkd. > For such system component it's important to be able to free memory > quickly and efficiently. Unfortunately the time process takes to free > up its memory after receiving a SIGKILL might vary based on the state > of the process (uninterruptible sleep), size and OPP level of the core > the process is running. A mechanism to free resources of the target > process in a more predictable way would improve system's ability to > control its memory pressure. > Introduce process_mrelease system call that releases memory of a dying > process from the context of the caller. This way the memory is freed in > a more controllable way with CPU affinity and priority of the caller. > The workload of freeing the memory will also be charged to the caller. > The operation is allowed only on a dying process. > > After previous discussions [1, 2, 3] the decision was made [4] to introduce > a dedicated system call to cover this use case. > > The API is as follows, > > int process_mrelease(int pidfd, unsigned int flags); > > DESCRIPTION > The process_mrelease() system call is used to free the memory of > an exiting process. > > The pidfd selects the process referred to by the PID file > descriptor. > (See pidofd_open(2) for further information) > > The flags argument is reserved for future use; currently, this > argument must be specified as 0. > > RETURN VALUE > On success, process_mrelease() returns 0. On error, -1 is > returned and errno is set to indicate the error. > > ERRORS > EBADF pidfd is not a valid PID file descriptor. > > EAGAIN Failed to release part of the address space. > > EINTR The call was interrupted by a signal; see signal(7). > > EINVAL flags is not 0. > > EINVAL The memory of the task cannot be released because the > process is not exiting, the address space is shared > with another live process or there is a core dump in > progress. > > ENOSYS This system call is not supported, for example, without > MMU support built into Linux. > > ESRCH The target process does not exist (i.e., it has terminated > and been waited on). > > [1] https://lore.kernel.org/lkml/20190411014353.113252-3-surenb@google.com/ > [2] https://lore.kernel.org/linux-api/20201113173448.1863419-1-surenb@google.com/ > [3] https://lore.kernel.org/linux-api/20201124053943.1684874-3-surenb@google.com/ > [4] https://lore.kernel.org/linux-api/20201223075712.GA4719@lst.de/ > > Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Shakeel Butt <shakeelb@google.com> Next I wanna see cgroup.procs giving me pidfds.
On Wed 04-08-21 11:50:03, Suren Baghdasaryan wrote: [...] > +SYSCALL_DEFINE2(process_mrelease, int, pidfd, unsigned int, flags) > +{ > +#ifdef CONFIG_MMU > + struct mm_struct *mm = NULL; > + struct task_struct *task; > + unsigned int f_flags; > + struct pid *pid; > + long ret = 0; > + > + if (flags) > + return -EINVAL; > + > + pid = pidfd_get_pid(pidfd, &f_flags); > + if (IS_ERR(pid)) > + return PTR_ERR(pid); > + > + task = get_pid_task(pid, PIDTYPE_PID); > + if (!task) { > + ret = -ESRCH; > + goto put_pid; > + } > + > + /* > + * If the task is dying and in the process of releasing its memory > + * then get its mm. > + */ > + task = find_lock_task_mm(task); You want a different task_struct because the returned one might be different from the given one and you already hold a reference which you do not want to leak > + if (!task) { > + ret = -ESRCH; > + goto put_pid; > + } > + if (task_will_free_mem(task) && (task->flags & PF_KTHREAD) == 0) { > + mm = task->mm; > + mmget(mm); > + } > + task_unlock(task); > + if (!mm) { > + ret = -EINVAL; > + goto put_task; > + } > + > + if (test_bit(MMF_OOM_SKIP, &mm->flags)) > + goto put_mm; This is too late to check for MMF_OOM_SKIP. task_will_free_mem will fail with the flag being set. I believe you want something like the following: p = find_lock_task_mm(task); mm = p->mm; /* The work has been done already */ if (test_bit(MMF_OOM_SKIP, &mm->flags)) { task_unlock(p); goto put_task; } i if (!task_will_free_mem(p)) { task_unlock(p); goto put_task; } mmget(mm); task_unlock(p); > + > + if (mmap_read_lock_killable(mm)) { > + ret = -EINTR; > + goto put_mm; > + } > + if (!__oom_reap_task_mm(mm)) > + ret = -EAGAIN; > + mmap_read_unlock(mm); > + > +put_mm: > + mmput(mm); > +put_task: > + put_task_struct(task); > +put_pid: > + put_pid(pid); > + return ret; > +#else > + return -ENOSYS; > +#endif /* CONFIG_MMU */ > +} > -- > 2.32.0.554.ge1b32706d8-goog
On Thu, Aug 5, 2021 at 12:10 AM Michal Hocko <mhocko@suse.com> wrote: > > On Wed 04-08-21 11:50:03, Suren Baghdasaryan wrote: > [...] > > +SYSCALL_DEFINE2(process_mrelease, int, pidfd, unsigned int, flags) > > +{ > > +#ifdef CONFIG_MMU > > + struct mm_struct *mm = NULL; > > + struct task_struct *task; > > + unsigned int f_flags; > > + struct pid *pid; > > + long ret = 0; > > + > > + if (flags) > > + return -EINVAL; > > + > > + pid = pidfd_get_pid(pidfd, &f_flags); > > + if (IS_ERR(pid)) > > + return PTR_ERR(pid); > > + > > + task = get_pid_task(pid, PIDTYPE_PID); > > + if (!task) { > > + ret = -ESRCH; > > + goto put_pid; > > + } > > + > > + /* > > + * If the task is dying and in the process of releasing its memory > > + * then get its mm. > > + */ > > + task = find_lock_task_mm(task); > > You want a different task_struct because the returned one might be > different from the given one and you already hold a reference which you > do not want to leak Ah, right. I was looking at the task locking and find_lock_task_mm() handles that but I missed the task pinning part. Will fix. > > > + if (!task) { > > + ret = -ESRCH; > > + goto put_pid; > > + } > > + if (task_will_free_mem(task) && (task->flags & PF_KTHREAD) == 0) { > > + mm = task->mm; > > + mmget(mm); > > + } > > + task_unlock(task); > > + if (!mm) { > > + ret = -EINVAL; > > + goto put_task; > > + } > > + > > + if (test_bit(MMF_OOM_SKIP, &mm->flags)) > > + goto put_mm; > > This is too late to check for MMF_OOM_SKIP. task_will_free_mem will fail > with the flag being set. I believe you want something like the > following: > > p = find_lock_task_mm(task); > mm = p->mm; > > /* The work has been done already */ > if (test_bit(MMF_OOM_SKIP, &mm->flags)) { > task_unlock(p); > goto put_task; > } > > i > if (!task_will_free_mem(p)) { > task_unlock(p); > goto put_task; > } > > mmget(mm); > task_unlock(p); > I see. Let me update the patch and will ask Andrew to remove the previous version from mm tree. Thanks for reviewing and pointing out the issues! > > > + > > + if (mmap_read_lock_killable(mm)) { > > + ret = -EINTR; > > + goto put_mm; > > + } > > + if (!__oom_reap_task_mm(mm)) > > + ret = -EAGAIN; > > + mmap_read_unlock(mm); > > + > > +put_mm: > > + mmput(mm); > > +put_task: > > + put_task_struct(task); > > +put_pid: > > + put_pid(pid); > > + return ret; > > +#else > > + return -ENOSYS; > > +#endif /* CONFIG_MMU */ > > +} > > -- > > 2.32.0.554.ge1b32706d8-goog > > -- > Michal Hocko > SUSE Labs
> > I see. Let me update the patch and will ask Andrew to remove the > previous version from mm tree. No need to ask. Just resend and Andrew will (usually) replace the old version automatically :)
On Thu, Aug 5, 2021 at 8:31 AM David Hildenbrand <david@redhat.com> wrote: > > > > > I see. Let me update the patch and will ask Andrew to remove the > > previous version from mm tree. > > No need to ask. Just resend and Andrew will (usually) replace the old > version automatically :) Done: https://lore.kernel.org/linux-mm/20210805170859.2389276-1-surenb@google.com Thanks! > > > -- > Thanks, > > David / dhildenb >
diff --git a/mm/oom_kill.c b/mm/oom_kill.c index c729a4c4a1ac..4f43ee79f663 100644 --- a/mm/oom_kill.c +++ b/mm/oom_kill.c @@ -28,6 +28,7 @@ #include <linux/sched/task.h> #include <linux/sched/debug.h> #include <linux/swap.h> +#include <linux/syscalls.h> #include <linux/timex.h> #include <linux/jiffies.h> #include <linux/cpuset.h> @@ -1141,3 +1142,67 @@ void pagefault_out_of_memory(void) out_of_memory(&oc); mutex_unlock(&oom_lock); } + +SYSCALL_DEFINE2(process_mrelease, int, pidfd, unsigned int, flags) +{ +#ifdef CONFIG_MMU + struct mm_struct *mm = NULL; + struct task_struct *task; + unsigned int f_flags; + struct pid *pid; + long ret = 0; + + if (flags) + return -EINVAL; + + pid = pidfd_get_pid(pidfd, &f_flags); + if (IS_ERR(pid)) + return PTR_ERR(pid); + + task = get_pid_task(pid, PIDTYPE_PID); + if (!task) { + ret = -ESRCH; + goto put_pid; + } + + /* + * If the task is dying and in the process of releasing its memory + * then get its mm. + */ + task = find_lock_task_mm(task); + if (!task) { + ret = -ESRCH; + goto put_pid; + } + if (task_will_free_mem(task) && (task->flags & PF_KTHREAD) == 0) { + mm = task->mm; + mmget(mm); + } + task_unlock(task); + if (!mm) { + ret = -EINVAL; + goto put_task; + } + + if (test_bit(MMF_OOM_SKIP, &mm->flags)) + goto put_mm; + + if (mmap_read_lock_killable(mm)) { + ret = -EINTR; + goto put_mm; + } + if (!__oom_reap_task_mm(mm)) + ret = -EAGAIN; + mmap_read_unlock(mm); + +put_mm: + mmput(mm); +put_task: + put_task_struct(task); +put_pid: + put_pid(pid); + return ret; +#else + return -ENOSYS; +#endif /* CONFIG_MMU */ +}
In modern systems it's not unusual to have a system component monitoring memory conditions of the system and tasked with keeping system memory pressure under control. One way to accomplish that is to kill non-essential processes to free up memory for more important ones. Examples of this are Facebook's OOM killer daemon called oomd and Android's low memory killer daemon called lmkd. For such system component it's important to be able to free memory quickly and efficiently. Unfortunately the time process takes to free up its memory after receiving a SIGKILL might vary based on the state of the process (uninterruptible sleep), size and OPP level of the core the process is running. A mechanism to free resources of the target process in a more predictable way would improve system's ability to control its memory pressure. Introduce process_mrelease system call that releases memory of a dying process from the context of the caller. This way the memory is freed in a more controllable way with CPU affinity and priority of the caller. The workload of freeing the memory will also be charged to the caller. The operation is allowed only on a dying process. After previous discussions [1, 2, 3] the decision was made [4] to introduce a dedicated system call to cover this use case. The API is as follows, int process_mrelease(int pidfd, unsigned int flags); DESCRIPTION The process_mrelease() system call is used to free the memory of an exiting process. The pidfd selects the process referred to by the PID file descriptor. (See pidofd_open(2) for further information) The flags argument is reserved for future use; currently, this argument must be specified as 0. RETURN VALUE On success, process_mrelease() returns 0. On error, -1 is returned and errno is set to indicate the error. ERRORS EBADF pidfd is not a valid PID file descriptor. EAGAIN Failed to release part of the address space. EINTR The call was interrupted by a signal; see signal(7). EINVAL flags is not 0. EINVAL The memory of the task cannot be released because the process is not exiting, the address space is shared with another live process or there is a core dump in progress. ENOSYS This system call is not supported, for example, without MMU support built into Linux. ESRCH The target process does not exist (i.e., it has terminated and been waited on). [1] https://lore.kernel.org/lkml/20190411014353.113252-3-surenb@google.com/ [2] https://lore.kernel.org/linux-api/20201113173448.1863419-1-surenb@google.com/ [3] https://lore.kernel.org/linux-api/20201124053943.1684874-3-surenb@google.com/ [4] https://lore.kernel.org/linux-api/20201223075712.GA4719@lst.de/ Signed-off-by: Suren Baghdasaryan <surenb@google.com> --- changes in v5: - Changed links, per David Hildenbrand and Michal Hocko - Condensed the background section in the description, per David Hildenbrand - Changed flags check, per David Hildenbrand - Changed description for ENOSYS in the manual pages, per David Hildenbrand - Changed wording from "SIGKILLed" to "exiting" process in the manual pages description, per Michal Hocko - Used find_lock_task_mm() to lock the task_struct, per Michal Hocko - Added check for MMF_OOM_SKIP, per Michal Hocko changes in v6: - Changed MMF_OOM_SKIP handling by returning success, per Michal Hocko mm/oom_kill.c | 65 +++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 65 insertions(+)