Donate to e Foundation | Murena handsets with /e/OS | Own a part of Murena! Learn more

Commit 724adb44 authored by Charan Teja Reddy's avatar Charan Teja Reddy
Browse files

mm: oom_kill: reap memory of a task that receives SIGKILL 



Free the pages parallely for a task that receives SIGKILL using the
oom_reaper. This freeing of pages will help to give the pages to buddy
system well advance.
This reaps for the process which received SIGKILL through
either sys_kill from user or kill_pid from kernel and that sending
process has CAP_KILL capability.
Also sysctl interface, reap_mem_on_sigkill, is added to turn on/off this
feature.

Change-Id: I21adb95de5e380a80d7eb0b87d9b5b553f52e28a
Signed-off-by: default avatarCharan Teja Reddy <charante@codeaurora.org>
parent ada19597

Documentation/sysctl/vm.txt

+19 −0
Original line number Diff line number Diff line
@@ -49,6 +49,7 @@ Currently, these files are in /proc/sys/vm: 
- nr_trim_pages         (only if CONFIG_MMU=n)

- numa_zonelist_order

- oom_dump_tasks

- reap_mem_on_sigkill

- oom_kill_allocating_task

- overcommit_kbytes

- overcommit_memory
@@ -640,6 +641,24 @@ The default value is 1 (enabled). 


==============================================================



reap_mem_on_sigkill



This enables or disables the memory reaping for a SIGKILL received

process and that the sending process must have the CAP_KILL capabilities.



If this is set to 1, when a process receives SIGKILL from a process

that has the capability, CAP_KILL, the process is added into the oom_reaper

queue which can be picked up by the oom_reaper thread to reap the memory of

that process. This reaps for the process which received SIGKILL through

either sys_kill from user or kill_pid from kernel.



If this is set to 0, we are not reaping memory of a SIGKILL, sent through

either sys_kill from user or kill_pid from kernel, received process.



The default value is 0 (disabled).



==============================================================



oom_kill_allocating_task



This enables or disables killing the OOM-triggering task in

include/linux/oom.h

+4 −0
Original line number Diff line number Diff line
@@ -120,4 +120,8 @@ extern void dump_tasks(struct mem_cgroup *memcg, 
extern int sysctl_oom_dump_tasks;

extern int sysctl_oom_kill_allocating_task;

extern int sysctl_panic_on_oom;

extern int sysctl_reap_mem_on_sigkill;



/* calls for LMK reaper */

extern void add_to_oom_reaper(struct task_struct *p);

#endif /* _INCLUDE_LINUX_OOM_H */

kernel/signal.c

+6 −1
Original line number Diff line number Diff line
@@ -40,6 +40,8 @@ 
#include <linux/cn_proc.h>

#include <linux/compiler.h>

#include <linux/posix-timers.h>

#include <linux/oom.h>

#include <linux/capability.h>



#define CREATE_TRACE_POINTS

#include <trace/events/signal.h>
@@ -1284,8 +1286,11 @@ int group_send_sig_info(int sig, struct siginfo *info, struct task_struct *p) 
	ret = check_kill_permission(sig, info, p);

	rcu_read_unlock();



	if (!ret && sig)

	if (!ret && sig) {

		ret = do_send_sig_info(sig, info, p, true);

		if (capable(CAP_KILL) && sig == SIGKILL)

			add_to_oom_reaper(p);

	}



	return ret;

}

kernel/sysctl.c

+7 −0
Original line number Diff line number Diff line
@@ -1377,6 +1377,13 @@ static struct ctl_table vm_table[] = { 
		.mode		= 0644,

		.proc_handler	= proc_dointvec,

	},

	{

		.procname       = "reap_mem_on_sigkill",

		.data           = &sysctl_reap_mem_on_sigkill,

		.maxlen         = sizeof(sysctl_reap_mem_on_sigkill),

		.mode           = 0644,

		.proc_handler   = proc_dointvec,

	},

	{

		.procname	= "overcommit_ratio",

		.data		= &sysctl_overcommit_ratio,

mm/oom_kill.c

+40 −8
Original line number Diff line number Diff line
@@ -51,6 +51,7 @@ 
int sysctl_panic_on_oom;

int sysctl_oom_kill_allocating_task;

int sysctl_oom_dump_tasks = 1;

int sysctl_reap_mem_on_sigkill;



DEFINE_MUTEX(oom_lock);
@@ -628,13 +629,21 @@ static void wake_oom_reaper(struct task_struct *tsk) 
	if (!oom_reaper_th)

		return;



	/*

	 * Move the lock here to avoid scenario of queuing

	 * the same task by both OOM killer and any other SIGKILL

	 * path.

	 */

	spin_lock(&oom_reaper_lock);



	/* tsk is already queued? */

	if (tsk == oom_reaper_list || tsk->oom_reaper_list)

	if (tsk == oom_reaper_list || tsk->oom_reaper_list) {

		spin_unlock(&oom_reaper_lock);

		return;

	}



	get_task_struct(tsk);



	spin_lock(&oom_reaper_lock);

	tsk->oom_reaper_list = oom_reaper_list;

	oom_reaper_list = tsk;

	spin_unlock(&oom_reaper_lock);
@@ -659,6 +668,16 @@ static inline void wake_oom_reaper(struct task_struct *tsk) 
}

#endif /* CONFIG_MMU */



static void __mark_oom_victim(struct task_struct *tsk)

{

	struct mm_struct *mm = tsk->mm;



	if (!cmpxchg(&tsk->signal->oom_mm, NULL, mm)) {

		mmgrab(tsk->signal->oom_mm);

		set_bit(MMF_OOM_VICTIM, &mm->flags);

	}

}



/**

 * mark_oom_victim - mark the given task as OOM victim

 * @tsk: task to mark
@@ -671,18 +690,13 @@ static inline void wake_oom_reaper(struct task_struct *tsk) 
 */

static void mark_oom_victim(struct task_struct *tsk)

{

	struct mm_struct *mm = tsk->mm;



	WARN_ON(oom_killer_disabled);

	/* OOM killer might race with memcg OOM */

	if (test_and_set_tsk_thread_flag(tsk, TIF_MEMDIE))

		return;



	/* oom_mm is bound to the signal struct life time. */

	if (!cmpxchg(&tsk->signal->oom_mm, NULL, mm)) {

		mmgrab(tsk->signal->oom_mm);

		set_bit(MMF_OOM_VICTIM, &mm->flags);

	}

	__mark_oom_victim(tsk);



	/*

	 * Make sure that the task is woken up from uninterruptible sleep
@@ -1106,3 +1120,21 @@ void pagefault_out_of_memory(void) 
	out_of_memory(&oc);

	mutex_unlock(&oom_lock);

}



void add_to_oom_reaper(struct task_struct *p)

{

	if (!sysctl_reap_mem_on_sigkill)

		return;



	p = find_lock_task_mm(p);

	if (!p)

		return;



	get_task_struct(p);

	if (task_will_free_mem(p)) {

		__mark_oom_victim(p);

		wake_oom_reaper(p);

	}

	task_unlock(p);

	put_task_struct(p);

}