[net-next-2.6.git] / arch / powerpc / platforms / pseries / cmm.c

/*
 * Collaborative memory management interface.
 *
 * Copyright (C) 2008 IBM Corporation
 * Author(s): Brian King (brking@linux.vnet.ibm.com),
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 *
 */

#include <linux/ctype.h>
#include <linux/delay.h>
#include <linux/errno.h>
#include <linux/fs.h>
#include <linux/gfp.h>
#include <linux/init.h>
#include <linux/kthread.h>
#include <linux/module.h>
#include <linux/oom.h>
#include <linux/reboot.h>
#include <linux/sched.h>
#include <linux/stringify.h>
#include <linux/swap.h>
#include <linux/sysdev.h>
#include <asm/firmware.h>
#include <asm/hvcall.h>
#include <asm/mmu.h>
#include <asm/pgalloc.h>
#include <asm/uaccess.h>
#include <linux/memory.h>

#include "plpar_wrappers.h"

#define CMM_DRIVER_VERSION	"1.0.0"
#define CMM_DEFAULT_DELAY	1
#define CMM_HOTPLUG_DELAY	5
#define CMM_DEBUG			0
#define CMM_DISABLE		0
#define CMM_OOM_KB		1024
#define CMM_MIN_MEM_MB		256
#define KB2PAGES(_p)		((_p)>>(PAGE_SHIFT-10))
#define PAGES2KB(_p)		((_p)<<(PAGE_SHIFT-10))
/*
 * The priority level tries to ensure that this notifier is called as
 * late as possible to reduce thrashing in the shared memory pool.
 */
#define CMM_MEM_HOTPLUG_PRI	1
#define CMM_MEM_ISOLATE_PRI	15

static unsigned int delay = CMM_DEFAULT_DELAY;
static unsigned int hotplug_delay = CMM_HOTPLUG_DELAY;
static unsigned int oom_kb = CMM_OOM_KB;
static unsigned int cmm_debug = CMM_DEBUG;
static unsigned int cmm_disabled = CMM_DISABLE;
static unsigned long min_mem_mb = CMM_MIN_MEM_MB;
static struct sys_device cmm_sysdev;

MODULE_AUTHOR("Brian King <brking@linux.vnet.ibm.com>");
MODULE_DESCRIPTION("IBM System p Collaborative Memory Manager");
MODULE_LICENSE("GPL");
MODULE_VERSION(CMM_DRIVER_VERSION);

module_param_named(delay, delay, uint, S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(delay, "Delay (in seconds) between polls to query hypervisor paging requests. "
		 "[Default=" __stringify(CMM_DEFAULT_DELAY) "]");
module_param_named(hotplug_delay, hotplug_delay, uint, S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(delay, "Delay (in seconds) after memory hotplug remove "
		 "before loaning resumes. "
		 "[Default=" __stringify(CMM_HOTPLUG_DELAY) "]");
module_param_named(oom_kb, oom_kb, uint, S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(oom_kb, "Amount of memory in kb to free on OOM. "
		 "[Default=" __stringify(CMM_OOM_KB) "]");
module_param_named(min_mem_mb, min_mem_mb, ulong, S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(min_mem_mb, "Minimum amount of memory (in MB) to not balloon. "
		 "[Default=" __stringify(CMM_MIN_MEM_MB) "]");
module_param_named(debug, cmm_debug, uint, S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(debug, "Enable module debugging logging. Set to 1 to enable. "
		 "[Default=" __stringify(CMM_DEBUG) "]");

#define CMM_NR_PAGES ((PAGE_SIZE - sizeof(void *) - sizeof(unsigned long)) / sizeof(unsigned long))

#define cmm_dbg(...) if (cmm_debug) { printk(KERN_INFO "cmm: "__VA_ARGS__); }

struct cmm_page_array {
	struct cmm_page_array *next;
	unsigned long index;
	unsigned long page[CMM_NR_PAGES];
};

static unsigned long loaned_pages;
static unsigned long loaned_pages_target;
static unsigned long oom_freed_pages;

static struct cmm_page_array *cmm_page_list;
static DEFINE_SPINLOCK(cmm_lock);

static DEFINE_MUTEX(hotplug_mutex);
static int hotplug_occurred; /* protected by the hotplug mutex */

static struct task_struct *cmm_thread_ptr;

/**
 * cmm_alloc_pages - Allocate pages and mark them as loaned
 * @nr:	number of pages to allocate
 *
 * Return value:
 * 	number of pages requested to be allocated which were not
 **/
static long cmm_alloc_pages(long nr)
{
	struct cmm_page_array *pa, *npa;
	unsigned long addr;
	long rc;

	cmm_dbg("Begin request for %ld pages\n", nr);

	while (nr) {
		/* Exit if a hotplug operation is in progress or occurred */
		if (mutex_trylock(&hotplug_mutex)) {
			if (hotplug_occurred) {
				mutex_unlock(&hotplug_mutex);
				break;
			}
			mutex_unlock(&hotplug_mutex);
		} else {
			break;
		}

		addr = __get_free_page(GFP_NOIO | __GFP_NOWARN |
				       __GFP_NORETRY | __GFP_NOMEMALLOC);
		if (!addr)
			break;
		spin_lock(&cmm_lock);
		pa = cmm_page_list;
		if (!pa || pa->index >= CMM_NR_PAGES) {
			/* Need a new page for the page list. */
			spin_unlock(&cmm_lock);
			npa = (struct cmm_page_array *)__get_free_page(
					GFP_NOIO | __GFP_NOWARN |
					__GFP_NORETRY | __GFP_NOMEMALLOC);
			if (!npa) {
				pr_info("%s: Can not allocate new page list\n", __func__);
				free_page(addr);
				break;
			}
			spin_lock(&cmm_lock);
			pa = cmm_page_list;

			if (!pa || pa->index >= CMM_NR_PAGES) {
				npa->next = pa;
				npa->index = 0;
				pa = npa;
				cmm_page_list = pa;
			} else
				free_page((unsigned long) npa);
		}

		if ((rc = plpar_page_set_loaned(__pa(addr)))) {
			pr_err("%s: Can not set page to loaned. rc=%ld\n", __func__, rc);
			spin_unlock(&cmm_lock);
			free_page(addr);
			break;
		}

		pa->page[pa->index++] = addr;
		loaned_pages++;
		totalram_pages--;
		spin_unlock(&cmm_lock);
		nr--;
	}

	cmm_dbg("End request with %ld pages unfulfilled\n", nr);
	return nr;
}

/**
 * cmm_free_pages - Free pages and mark them as active
 * @nr:	number of pages to free
 *
 * Return value:
 * 	number of pages requested to be freed which were not
 **/
static long cmm_free_pages(long nr)
{
	struct cmm_page_array *pa;
	unsigned long addr;

	cmm_dbg("Begin free of %ld pages.\n", nr);
	spin_lock(&cmm_lock);
	pa = cmm_page_list;
	while (nr) {
		if (!pa || pa->index <= 0)
			break;
		addr = pa->page[--pa->index];

		if (pa->index == 0) {
			pa = pa->next;
			free_page((unsigned long) cmm_page_list);
			cmm_page_list = pa;
		}

		plpar_page_set_active(__pa(addr));
		free_page(addr);
		loaned_pages--;
		nr--;
		totalram_pages++;
	}
	spin_unlock(&cmm_lock);
	cmm_dbg("End request with %ld pages unfulfilled\n", nr);
	return nr;
}

/**
 * cmm_oom_notify - OOM notifier
 * @self:	notifier block struct
 * @dummy:	not used
 * @parm:	returned - number of pages freed
 *
 * Return value:
 * 	NOTIFY_OK
 **/
static int cmm_oom_notify(struct notifier_block *self,
			  unsigned long dummy, void *parm)
{
	unsigned long *freed = parm;
	long nr = KB2PAGES(oom_kb);

	cmm_dbg("OOM processing started\n");
	nr = cmm_free_pages(nr);
	loaned_pages_target = loaned_pages;
	*freed += KB2PAGES(oom_kb) - nr;
	oom_freed_pages += KB2PAGES(oom_kb) - nr;
	cmm_dbg("OOM processing complete\n");
	return NOTIFY_OK;
}

/**
 * cmm_get_mpp - Read memory performance parameters
 *
 * Makes hcall to query the current page loan request from the hypervisor.
 *
 * Return value:
 * 	nothing
 **/
static void cmm_get_mpp(void)
{
	int rc;
	struct hvcall_mpp_data mpp_data;
	signed long active_pages_target, page_loan_request, target;
	signed long total_pages = totalram_pages + loaned_pages;
	signed long min_mem_pages = (min_mem_mb * 1024 * 1024) / PAGE_SIZE;

	rc = h_get_mpp(&mpp_data);

	if (rc != H_SUCCESS)
		return;

	page_loan_request = div_s64((s64)mpp_data.loan_request, PAGE_SIZE);
	target = page_loan_request + (signed long)loaned_pages;

	if (target < 0 || total_pages < min_mem_pages)
		target = 0;

	if (target > oom_freed_pages)
		target -= oom_freed_pages;
	else
		target = 0;

	active_pages_target = total_pages - target;

	if (min_mem_pages > active_pages_target)
		target = total_pages - min_mem_pages;

	if (target < 0)
		target = 0;

	loaned_pages_target = target;

	cmm_dbg("delta = %ld, loaned = %lu, target = %lu, oom = %lu, totalram = %lu\n",
		page_loan_request, loaned_pages, loaned_pages_target,
		oom_freed_pages, totalram_pages);
}

static struct notifier_block cmm_oom_nb = {
	.notifier_call = cmm_oom_notify
};

/**
 * cmm_thread - CMM task thread
 * @dummy:	not used
 *
 * Return value:
 * 	0
 **/
static int cmm_thread(void *dummy)
{
	unsigned long timeleft;

	while (1) {
		timeleft = msleep_interruptible(delay * 1000);

		if (kthread_should_stop() || timeleft)
			break;

		if (mutex_trylock(&hotplug_mutex)) {
			if (hotplug_occurred) {
				hotplug_occurred = 0;
				mutex_unlock(&hotplug_mutex);
				cmm_dbg("Hotplug operation has occurred, "
						"loaning activity suspended "
						"for %d seconds.\n",
						hotplug_delay);
				timeleft = msleep_interruptible(hotplug_delay *
						1000);
				if (kthread_should_stop() || timeleft)
					break;
				continue;
			}
			mutex_unlock(&hotplug_mutex);
		} else {
			cmm_dbg("Hotplug operation in progress, activity "
					"suspended\n");
			continue;
		}

		cmm_get_mpp();

		if (loaned_pages_target > loaned_pages) {
			if (cmm_alloc_pages(loaned_pages_target - loaned_pages))
				loaned_pages_target = loaned_pages;
		} else if (loaned_pages_target < loaned_pages)
			cmm_free_pages(loaned_pages - loaned_pages_target);
	}
	return 0;
}

#define CMM_SHOW(name, format, args...)			\
	static ssize_t show_##name(struct sys_device *dev,	\
				   struct sysdev_attribute *attr,	\
				   char *buf)			\
	{							\
		return sprintf(buf, format, ##args);		\
	}							\
	static SYSDEV_ATTR(name, S_IRUGO, show_##name, NULL)

CMM_SHOW(loaned_kb, "%lu\n", PAGES2KB(loaned_pages));
CMM_SHOW(loaned_target_kb, "%lu\n", PAGES2KB(loaned_pages_target));

static ssize_t show_oom_pages(struct sys_device *dev,
			      struct sysdev_attribute *attr, char *buf)
{
	return sprintf(buf, "%lu\n", PAGES2KB(oom_freed_pages));
}

static ssize_t store_oom_pages(struct sys_device *dev,
			       struct sysdev_attribute *attr,
			       const char *buf, size_t count)
{
	unsigned long val = simple_strtoul (buf, NULL, 10);

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
	if (val != 0)
		return -EBADMSG;

	oom_freed_pages = 0;
	return count;
}

static SYSDEV_ATTR(oom_freed_kb, S_IWUSR| S_IRUGO,
		   show_oom_pages, store_oom_pages);

static struct sysdev_attribute *cmm_attrs[] = {
	&attr_loaned_kb,
	&attr_loaned_target_kb,
	&attr_oom_freed_kb,
};

static struct sysdev_class cmm_sysdev_class = {
	.name = "cmm",
};

/**
 * cmm_sysfs_register - Register with sysfs
 *
 * Return value:
 * 	0 on success / other on failure
 **/
static int cmm_sysfs_register(struct sys_device *sysdev)
{
	int i, rc;

	if ((rc = sysdev_class_register(&cmm_sysdev_class)))
		return rc;

	sysdev->id = 0;
	sysdev->cls = &cmm_sysdev_class;

	if ((rc = sysdev_register(sysdev)))
		goto class_unregister;

	for (i = 0; i < ARRAY_SIZE(cmm_attrs); i++) {
		if ((rc = sysdev_create_file(sysdev, cmm_attrs[i])))
			goto fail;
	}

	return 0;

fail:
	while (--i >= 0)
		sysdev_remove_file(sysdev, cmm_attrs[i]);
	sysdev_unregister(sysdev);
class_unregister:
	sysdev_class_unregister(&cmm_sysdev_class);
	return rc;
}

/**
 * cmm_unregister_sysfs - Unregister from sysfs
 *
 **/
static void cmm_unregister_sysfs(struct sys_device *sysdev)
{
	int i;

	for (i = 0; i < ARRAY_SIZE(cmm_attrs); i++)
		sysdev_remove_file(sysdev, cmm_attrs[i]);
	sysdev_unregister(sysdev);
	sysdev_class_unregister(&cmm_sysdev_class);
}

/**
 * cmm_reboot_notifier - Make sure pages are not still marked as "loaned"
 *
 **/
static int cmm_reboot_notifier(struct notifier_block *nb,
			       unsigned long action, void *unused)
{
	if (action == SYS_RESTART) {
		if (cmm_thread_ptr)
			kthread_stop(cmm_thread_ptr);
		cmm_thread_ptr = NULL;
		cmm_free_pages(loaned_pages);
	}
	return NOTIFY_DONE;
}

static struct notifier_block cmm_reboot_nb = {
	.notifier_call = cmm_reboot_notifier,
};

/**
 * cmm_count_pages - Count the number of pages loaned in a particular range.
 *
 * @arg: memory_isolate_notify structure with address range and count
 *
 * Return value:
 *      0 on success
 **/
static unsigned long cmm_count_pages(void *arg)
{
	struct memory_isolate_notify *marg = arg;
	struct cmm_page_array *pa;
	unsigned long start = (unsigned long)pfn_to_kaddr(marg->start_pfn);
	unsigned long end = start + (marg->nr_pages << PAGE_SHIFT);
	unsigned long idx;

	spin_lock(&cmm_lock);
	pa = cmm_page_list;
	while (pa) {
		if ((unsigned long)pa >= start && (unsigned long)pa < end)
			marg->pages_found++;
		for (idx = 0; idx < pa->index; idx++)
			if (pa->page[idx] >= start && pa->page[idx] < end)
				marg->pages_found++;
		pa = pa->next;
	}
	spin_unlock(&cmm_lock);
	return 0;
}

/**
 * cmm_memory_isolate_cb - Handle memory isolation notifier calls
 * @self:	notifier block struct
 * @action:	action to take
 * @arg:	struct memory_isolate_notify data for handler
 *
 * Return value:
 *	NOTIFY_OK or notifier error based on subfunction return value
 **/
static int cmm_memory_isolate_cb(struct notifier_block *self,
				 unsigned long action, void *arg)
{
	int ret = 0;

	if (action == MEM_ISOLATE_COUNT)
		ret = cmm_count_pages(arg);

	if (ret)
		ret = notifier_from_errno(ret);
	else
		ret = NOTIFY_OK;

	return ret;
}

static struct notifier_block cmm_mem_isolate_nb = {
	.notifier_call = cmm_memory_isolate_cb,
	.priority = CMM_MEM_ISOLATE_PRI
};

/**
 * cmm_mem_going_offline - Unloan pages where memory is to be removed
 * @arg: memory_notify structure with page range to be offlined
 *
 * Return value:
 *	0 on success
 **/
static int cmm_mem_going_offline(void *arg)
{
	struct memory_notify *marg = arg;
	unsigned long start_page = (unsigned long)pfn_to_kaddr(marg->start_pfn);
	unsigned long end_page = start_page + (marg->nr_pages << PAGE_SHIFT);
	struct cmm_page_array *pa_curr, *pa_last, *npa;
	unsigned long idx;
	unsigned long freed = 0;

	cmm_dbg("Memory going offline, searching 0x%lx (%ld pages).\n",
			start_page, marg->nr_pages);
	spin_lock(&cmm_lock);

	/* Search the page list for pages in the range to be offlined */
	pa_last = pa_curr = cmm_page_list;
	while (pa_curr) {
		for (idx = (pa_curr->index - 1); (idx + 1) > 0; idx--) {
			if ((pa_curr->page[idx] < start_page) ||
			    (pa_curr->page[idx] >= end_page))
				continue;

			plpar_page_set_active(__pa(pa_curr->page[idx]));
			free_page(pa_curr->page[idx]);
			freed++;
			loaned_pages--;
			totalram_pages++;
			pa_curr->page[idx] = pa_last->page[--pa_last->index];
			if (pa_last->index == 0) {
				if (pa_curr == pa_last)
					pa_curr = pa_last->next;
				pa_last = pa_last->next;
				free_page((unsigned long)cmm_page_list);
				cmm_page_list = pa_last;
				continue;
			}
		}
		pa_curr = pa_curr->next;
	}

	/* Search for page list structures in the range to be offlined */
	pa_last = NULL;
	pa_curr = cmm_page_list;
	while (pa_curr) {
		if (((unsigned long)pa_curr >= start_page) &&
				((unsigned long)pa_curr < end_page)) {
			npa = (struct cmm_page_array *)__get_free_page(
					GFP_NOIO | __GFP_NOWARN |
					__GFP_NORETRY | __GFP_NOMEMALLOC);
			if (!npa) {
				spin_unlock(&cmm_lock);
				cmm_dbg("Failed to allocate memory for list "
						"management. Memory hotplug "
						"failed.\n");
				return ENOMEM;
			}
			memcpy(npa, pa_curr, PAGE_SIZE);
			if (pa_curr == cmm_page_list)
				cmm_page_list = npa;
			if (pa_last)
				pa_last->next = npa;
			free_page((unsigned long) pa_curr);
			freed++;
			pa_curr = npa;
		}

		pa_last = pa_curr;
		pa_curr = pa_curr->next;
	}

	spin_unlock(&cmm_lock);
	cmm_dbg("Released %ld pages in the search range.\n", freed);

	return 0;
}

/**
 * cmm_memory_cb - Handle memory hotplug notifier calls
 * @self:	notifier block struct
 * @action:	action to take
 * @arg:	struct memory_notify data for handler
 *
 * Return value:
 *	NOTIFY_OK or notifier error based on subfunction return value
 *
 **/
static int cmm_memory_cb(struct notifier_block *self,
			unsigned long action, void *arg)
{
	int ret = 0;

	switch (action) {
	case MEM_GOING_OFFLINE:
		mutex_lock(&hotplug_mutex);
		hotplug_occurred = 1;
		ret = cmm_mem_going_offline(arg);
		break;
	case MEM_OFFLINE:
	case MEM_CANCEL_OFFLINE:
		mutex_unlock(&hotplug_mutex);
		cmm_dbg("Memory offline operation complete.\n");
		break;
	case MEM_GOING_ONLINE:
	case MEM_ONLINE:
	case MEM_CANCEL_ONLINE:
		break;
	}

	if (ret)
		ret = notifier_from_errno(ret);
	else
		ret = NOTIFY_OK;

	return ret;
}

static struct notifier_block cmm_mem_nb = {
	.notifier_call = cmm_memory_cb,
	.priority = CMM_MEM_HOTPLUG_PRI
};

/**
 * cmm_init - Module initialization
 *
 * Return value:
 * 	0 on success / other on failure
 **/
static int cmm_init(void)
{
	int rc = -ENOMEM;

	if (!firmware_has_feature(FW_FEATURE_CMO))
		return -EOPNOTSUPP;

	if ((rc = register_oom_notifier(&cmm_oom_nb)) < 0)
		return rc;

	if ((rc = register_reboot_notifier(&cmm_reboot_nb)))
		goto out_oom_notifier;

	if ((rc = cmm_sysfs_register(&cmm_sysdev)))
		goto out_reboot_notifier;

	if (register_memory_notifier(&cmm_mem_nb) ||
	    register_memory_isolate_notifier(&cmm_mem_isolate_nb))
		goto out_unregister_notifier;

	if (cmm_disabled)
		return rc;

	cmm_thread_ptr = kthread_run(cmm_thread, NULL, "cmmthread");
	if (IS_ERR(cmm_thread_ptr)) {
		rc = PTR_ERR(cmm_thread_ptr);
		goto out_unregister_notifier;
	}

	return rc;

out_unregister_notifier:
	unregister_memory_notifier(&cmm_mem_nb);
	unregister_memory_isolate_notifier(&cmm_mem_isolate_nb);
	cmm_unregister_sysfs(&cmm_sysdev);
out_reboot_notifier:
	unregister_reboot_notifier(&cmm_reboot_nb);
out_oom_notifier:
	unregister_oom_notifier(&cmm_oom_nb);
	return rc;
}

/**
 * cmm_exit - Module exit
 *
 * Return value:
 * 	nothing
 **/
static void cmm_exit(void)
{
	if (cmm_thread_ptr)
		kthread_stop(cmm_thread_ptr);
	unregister_oom_notifier(&cmm_oom_nb);
	unregister_reboot_notifier(&cmm_reboot_nb);
	unregister_memory_notifier(&cmm_mem_nb);
	unregister_memory_isolate_notifier(&cmm_mem_isolate_nb);
	cmm_free_pages(loaned_pages);
	cmm_unregister_sysfs(&cmm_sysdev);
}

/**
 * cmm_set_disable - Disable/Enable CMM
 *
 * Return value:
 * 	0 on success / other on failure
 **/
static int cmm_set_disable(const char *val, struct kernel_param *kp)
{
	int disable = simple_strtoul(val, NULL, 10);

	if (disable != 0 && disable != 1)
		return -EINVAL;

	if (disable && !cmm_disabled) {
		if (cmm_thread_ptr)
			kthread_stop(cmm_thread_ptr);
		cmm_thread_ptr = NULL;
		cmm_free_pages(loaned_pages);
	} else if (!disable && cmm_disabled) {
		cmm_thread_ptr = kthread_run(cmm_thread, NULL, "cmmthread");
		if (IS_ERR(cmm_thread_ptr))
			return PTR_ERR(cmm_thread_ptr);
	}

	cmm_disabled = disable;
	return 0;
}

module_param_call(disable, cmm_set_disable, param_get_uint,
		  &cmm_disabled, S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(disable, "Disable CMM. Set to 1 to disable. "
		 "[Default=" __stringify(CMM_DISABLE) "]");

module_init(cmm_init);
module_exit(cmm_exit);
Commit	Line	Data
84af458b BK	1	/*
	2	* Collaborative memory management interface.
	3	*
	4	* Copyright (C) 2008 IBM Corporation
	5	* Author(s): Brian King (brking@linux.vnet.ibm.com),
	6	*
	7	* This program is free software; you can redistribute it and/or modify
	8	* it under the terms of the GNU General Public License as published by
	9	* the Free Software Foundation; either version 2 of the License, or
	10	* (at your option) any later version.
	11	*
	12	* This program is distributed in the hope that it will be useful,
	13	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	14	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	15	* GNU General Public License for more details.
	16	*
	17	* You should have received a copy of the GNU General Public License
	18	* along with this program; if not, write to the Free Software
	19	* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
	20	*
	21	*/
	22
	23	#include <linux/ctype.h>
	24	#include <linux/delay.h>
	25	#include <linux/errno.h>
	26	#include <linux/fs.h>
5a0e3ad6	27	#include <linux/gfp.h>
84af458b BK	28	#include <linux/init.h>
	29	#include <linux/kthread.h>
	30	#include <linux/module.h>
	31	#include <linux/oom.h>
fecba962	32	#include <linux/reboot.h>
84af458b BK	33	#include <linux/sched.h>
	34	#include <linux/stringify.h>
	35	#include <linux/swap.h>
	36	#include <linux/sysdev.h>
	37	#include <asm/firmware.h>
	38	#include <asm/hvcall.h>
	39	#include <asm/mmu.h>
	40	#include <asm/pgalloc.h>
	41	#include <asm/uaccess.h>
14b8a76b	42	#include <linux/memory.h>
84af458b BK	43
	44	#include "plpar_wrappers.h"
	45
	46	#define CMM_DRIVER_VERSION "1.0.0"
	47	#define CMM_DEFAULT_DELAY 1
14b8a76b	48	#define CMM_HOTPLUG_DELAY 5
84af458b BK	49	#define CMM_DEBUG 0
	50	#define CMM_DISABLE 0
	51	#define CMM_OOM_KB 1024
	52	#define CMM_MIN_MEM_MB 256
	53	#define KB2PAGES(_p) ((_p)>>(PAGE_SHIFT-10))
	54	#define PAGES2KB(_p) ((_p)<<(PAGE_SHIFT-10))
14b8a76b RJ	55	/*
	56	* The priority level tries to ensure that this notifier is called as
	57	* late as possible to reduce thrashing in the shared memory pool.
	58	*/
	59	#define CMM_MEM_HOTPLUG_PRI 1
	60	#define CMM_MEM_ISOLATE_PRI 15
84af458b BK	61
84af458b BK	62	static unsigned int delay = CMM_DEFAULT_DELAY;
14b8a76b	63	static unsigned int hotplug_delay = CMM_HOTPLUG_DELAY;
84af458b BK	64	static unsigned int oom_kb = CMM_OOM_KB;
	65	static unsigned int cmm_debug = CMM_DEBUG;
	66	static unsigned int cmm_disabled = CMM_DISABLE;
	67	static unsigned long min_mem_mb = CMM_MIN_MEM_MB;
	68	static struct sys_device cmm_sysdev;
	69
	70	MODULE_AUTHOR("Brian King <brking@linux.vnet.ibm.com>");
	71	MODULE_DESCRIPTION("IBM System p Collaborative Memory Manager");
	72	MODULE_LICENSE("GPL");
	73	MODULE_VERSION(CMM_DRIVER_VERSION);
	74
	75	module_param_named(delay, delay, uint, S_IRUGO \| S_IWUSR);
	76	MODULE_PARM_DESC(delay, "Delay (in seconds) between polls to query hypervisor paging requests. "
	77	"[Default=" __stringify(CMM_DEFAULT_DELAY) "]");
14b8a76b RJ	78	module_param_named(hotplug_delay, hotplug_delay, uint, S_IRUGO \| S_IWUSR);
	79	MODULE_PARM_DESC(delay, "Delay (in seconds) after memory hotplug remove "
	80	"before loaning resumes. "
	81	"[Default=" __stringify(CMM_HOTPLUG_DELAY) "]");
84af458b BK	82	module_param_named(oom_kb, oom_kb, uint, S_IRUGO \| S_IWUSR);
	83	MODULE_PARM_DESC(oom_kb, "Amount of memory in kb to free on OOM. "
	84	"[Default=" __stringify(CMM_OOM_KB) "]");
	85	module_param_named(min_mem_mb, min_mem_mb, ulong, S_IRUGO \| S_IWUSR);
	86	MODULE_PARM_DESC(min_mem_mb, "Minimum amount of memory (in MB) to not balloon. "
	87	"[Default=" __stringify(CMM_MIN_MEM_MB) "]");
	88	module_param_named(debug, cmm_debug, uint, S_IRUGO \| S_IWUSR);
	89	MODULE_PARM_DESC(debug, "Enable module debugging logging. Set to 1 to enable. "
	90	"[Default=" __stringify(CMM_DEBUG) "]");
	91
	92	#define CMM_NR_PAGES ((PAGE_SIZE - sizeof(void *) - sizeof(unsigned long)) / sizeof(unsigned long))
	93
	94	#define cmm_dbg(...) if (cmm_debug) { printk(KERN_INFO "cmm: "__VA_ARGS__); }
	95
	96	struct cmm_page_array {
	97	struct cmm_page_array *next;
	98	unsigned long index;
	99	unsigned long page[CMM_NR_PAGES];
	100	};
	101
	102	static unsigned long loaned_pages;
	103	static unsigned long loaned_pages_target;
	104	static unsigned long oom_freed_pages;
	105
	106	static struct cmm_page_array *cmm_page_list;
	107	static DEFINE_SPINLOCK(cmm_lock);
	108
14b8a76b RJ	109	static DEFINE_MUTEX(hotplug_mutex);
	110	static int hotplug_occurred; /* protected by the hotplug mutex */
	111
84af458b BK	112	static struct task_struct *cmm_thread_ptr;
	113
	114	/**
	115	* cmm_alloc_pages - Allocate pages and mark them as loaned
	116	* @nr: number of pages to allocate
	117	*
	118	* Return value:
	119	* number of pages requested to be allocated which were not
	120	**/
	121	static long cmm_alloc_pages(long nr)
	122	{
	123	struct cmm_page_array pa, npa;
	124	unsigned long addr;
	125	long rc;
	126
	127	cmm_dbg("Begin request for %ld pages\n", nr);
	128
	129	while (nr) {
14b8a76b RJ	130	/* Exit if a hotplug operation is in progress or occurred */
	131	if (mutex_trylock(&hotplug_mutex)) {
	132	if (hotplug_occurred) {
	133	mutex_unlock(&hotplug_mutex);
	134	break;
	135	}
	136	mutex_unlock(&hotplug_mutex);
	137	} else {
	138	break;
	139	}
	140
84af458b BK	141	addr = __get_free_page(GFP_NOIO \| __GFP_NOWARN \|
	142	__GFP_NORETRY \| __GFP_NOMEMALLOC);
	143	if (!addr)
	144	break;
	145	spin_lock(&cmm_lock);
	146	pa = cmm_page_list;
	147	if (!pa \|\| pa->index >= CMM_NR_PAGES) {
	148	/* Need a new page for the page list. */
	149	spin_unlock(&cmm_lock);
14b8a76b RJ	150	npa = (struct cmm_page_array *)__get_free_page(
	151	GFP_NOIO \| __GFP_NOWARN \|
	152	__GFP_NORETRY \| __GFP_NOMEMALLOC);
84af458b	153	if (!npa) {
5df72bf3	154	pr_info("%s: Can not allocate new page list\n", __func__);
84af458b BK	155	free_page(addr);
	156	break;
	157	}
	158	spin_lock(&cmm_lock);
	159	pa = cmm_page_list;
	160
	161	if (!pa \|\| pa->index >= CMM_NR_PAGES) {
	162	npa->next = pa;
	163	npa->index = 0;
	164	pa = npa;
	165	cmm_page_list = pa;
	166	} else
	167	free_page((unsigned long) npa);
	168	}
	169
	170	if ((rc = plpar_page_set_loaned(__pa(addr)))) {
5df72bf3	171	pr_err("%s: Can not set page to loaned. rc=%ld\n", __func__, rc);
84af458b BK	172	spin_unlock(&cmm_lock);
	173	free_page(addr);
	174	break;
	175	}
	176
	177	pa->page[pa->index++] = addr;
	178	loaned_pages++;
	179	totalram_pages--;
	180	spin_unlock(&cmm_lock);
	181	nr--;
	182	}
	183
	184	cmm_dbg("End request with %ld pages unfulfilled\n", nr);
	185	return nr;
	186	}
	187
	188	/**
	189	* cmm_free_pages - Free pages and mark them as active
	190	* @nr: number of pages to free
	191	*
	192	* Return value:
	193	* number of pages requested to be freed which were not
	194	**/
	195	static long cmm_free_pages(long nr)
	196	{
	197	struct cmm_page_array *pa;
	198	unsigned long addr;
	199
	200	cmm_dbg("Begin free of %ld pages.\n", nr);
	201	spin_lock(&cmm_lock);
	202	pa = cmm_page_list;
	203	while (nr) {
	204	if (!pa \|\| pa->index <= 0)
	205	break;
	206	addr = pa->page[--pa->index];
	207
	208	if (pa->index == 0) {
	209	pa = pa->next;
	210	free_page((unsigned long) cmm_page_list);
	211	cmm_page_list = pa;
	212	}
	213
	214	plpar_page_set_active(__pa(addr));
	215	free_page(addr);
	216	loaned_pages--;
	217	nr--;
	218	totalram_pages++;
	219	}
	220	spin_unlock(&cmm_lock);
	221	cmm_dbg("End request with %ld pages unfulfilled\n", nr);
	222	return nr;
	223	}
	224
	225	/**
	226	* cmm_oom_notify - OOM notifier
	227	* @self: notifier block struct
	228	* @dummy: not used
	229	* @parm: returned - number of pages freed
	230	*
	231	* Return value:
	232	* NOTIFY_OK
	233	**/
	234	static int cmm_oom_notify(struct notifier_block *self,
	235	unsigned long dummy, void *parm)
236	{
237	unsigned long *freed = parm;
238	long nr = KB2PAGES(oom_kb);
239
240	cmm_dbg("OOM processing started\n");
241	nr = cmm_free_pages(nr);
242	loaned_pages_target = loaned_pages;
243	*freed += KB2PAGES(oom_kb) - nr;
244	oom_freed_pages += KB2PAGES(oom_kb) - nr;
245	cmm_dbg("OOM processing complete\n");
246	return NOTIFY_OK;
247	}
248
249	/**
250	* cmm_get_mpp - Read memory performance parameters
251	*
252	* Makes hcall to query the current page loan request from the hypervisor.
253	*
254	* Return value:
255	* nothing
256	**/
257	static void cmm_get_mpp(void)
258	{
259	int rc;
260	struct hvcall_mpp_data mpp_data;
8be8cf5b BK	261	signed long active_pages_target, page_loan_request, target;
	262	signed long total_pages = totalram_pages + loaned_pages;
	263	signed long min_mem_pages = (min_mem_mb * 1024 * 1024) / PAGE_SIZE;
84af458b BK	264
	265	rc = h_get_mpp(&mpp_data);
	266
	267	if (rc != H_SUCCESS)
	268	return;
	269
	270	page_loan_request = div_s64((s64)mpp_data.loan_request, PAGE_SIZE);
8be8cf5b BK	271	target = page_loan_request + (signed long)loaned_pages;
	272
	273	if (target < 0 \|\| total_pages < min_mem_pages)
	274	target = 0;
	275
	276	if (target > oom_freed_pages)
	277	target -= oom_freed_pages;
84af458b	278	else
8be8cf5b BK	279	target = 0;
	280
	281	active_pages_target = total_pages - target;
	282
	283	if (min_mem_pages > active_pages_target)
	284	target = total_pages - min_mem_pages;
84af458b	285
8be8cf5b BK	286	if (target < 0)
8be8cf5b BK	287	target = 0;
84af458b	288
8be8cf5b	289	loaned_pages_target = target;
84af458b BK	290
	291	cmm_dbg("delta = %ld, loaned = %lu, target = %lu, oom = %lu, totalram = %lu\n",
	292	page_loan_request, loaned_pages, loaned_pages_target,
	293	oom_freed_pages, totalram_pages);
	294	}
	295
	296	static struct notifier_block cmm_oom_nb = {
	297	.notifier_call = cmm_oom_notify
	298	};
	299
	300	/**
	301	* cmm_thread - CMM task thread
	302	* @dummy: not used
	303	*
	304	* Return value:
	305	* 0
	306	**/
	307	static int cmm_thread(void *dummy)
	308	{
	309	unsigned long timeleft;
	310
	311	while (1) {
	312	timeleft = msleep_interruptible(delay * 1000);
	313
14b8a76b	314	if (kthread_should_stop() \|\| timeleft)
84af458b	315	break;
14b8a76b RJ	316
	317	if (mutex_trylock(&hotplug_mutex)) {
	318	if (hotplug_occurred) {
	319	hotplug_occurred = 0;
	320	mutex_unlock(&hotplug_mutex);
	321	cmm_dbg("Hotplug operation has occurred, "
	322	"loaning activity suspended "
	323	"for %d seconds.\n",
	324	hotplug_delay);
	325	timeleft = msleep_interruptible(hotplug_delay *
	326	1000);
	327	if (kthread_should_stop() \|\| timeleft)
	328	break;
	329	continue;
	330	}
	331	mutex_unlock(&hotplug_mutex);
	332	} else {
	333	cmm_dbg("Hotplug operation in progress, activity "
	334	"suspended\n");
	335	continue;
84af458b BK	336	}
	337
	338	cmm_get_mpp();
	339
	340	if (loaned_pages_target > loaned_pages) {
	341	if (cmm_alloc_pages(loaned_pages_target - loaned_pages))
	342	loaned_pages_target = loaned_pages;
	343	} else if (loaned_pages_target < loaned_pages)
	344	cmm_free_pages(loaned_pages - loaned_pages_target);
	345	}
	346	return 0;
	347	}
	348
	349	#define CMM_SHOW(name, format, args...) \
3cee67f7 SR	350	static ssize_t show_##name(struct sys_device *dev, \
	351	struct sysdev_attribute *attr, \
	352	char *buf) \
84af458b BK	353	{ \
	354	return sprintf(buf, format, ##args); \
	355	} \
	356	static SYSDEV_ATTR(name, S_IRUGO, show_##name, NULL)
	357
	358	CMM_SHOW(loaned_kb, "%lu\n", PAGES2KB(loaned_pages));
	359	CMM_SHOW(loaned_target_kb, "%lu\n", PAGES2KB(loaned_pages_target));
	360
3cee67f7 SR	361	static ssize_t show_oom_pages(struct sys_device *dev,
3cee67f7 SR	362	struct sysdev_attribute attr, char buf)
84af458b BK	363	{
	364	return sprintf(buf, "%lu\n", PAGES2KB(oom_freed_pages));
	365	}
	366
	367	static ssize_t store_oom_pages(struct sys_device *dev,
3cee67f7	368	struct sysdev_attribute *attr,
84af458b BK	369	const char *buf, size_t count)
	370	{
	371	unsigned long val = simple_strtoul (buf, NULL, 10);
	372
	373	if (!capable(CAP_SYS_ADMIN))
	374	return -EPERM;
	375	if (val != 0)
	376	return -EBADMSG;
	377
	378	oom_freed_pages = 0;
	379	return count;
	380	}
	381
	382	static SYSDEV_ATTR(oom_freed_kb, S_IWUSR\| S_IRUGO,
	383	show_oom_pages, store_oom_pages);
	384
	385	static struct sysdev_attribute *cmm_attrs[] = {
	386	&attr_loaned_kb,
	387	&attr_loaned_target_kb,
	388	&attr_oom_freed_kb,
	389	};
	390
	391	static struct sysdev_class cmm_sysdev_class = {
	392	.name = "cmm",
	393	};
	394
	395	/**
	396	* cmm_sysfs_register - Register with sysfs
	397	*
	398	* Return value:
	399	* 0 on success / other on failure
	400	**/
	401	static int cmm_sysfs_register(struct sys_device *sysdev)
	402	{
	403	int i, rc;
	404
	405	if ((rc = sysdev_class_register(&cmm_sysdev_class)))
	406	return rc;
	407
	408	sysdev->id = 0;
	409	sysdev->cls = &cmm_sysdev_class;
	410
	411	if ((rc = sysdev_register(sysdev)))
	412	goto class_unregister;
	413
	414	for (i = 0; i < ARRAY_SIZE(cmm_attrs); i++) {
	415	if ((rc = sysdev_create_file(sysdev, cmm_attrs[i])))
	416	goto fail;
	417	}
	418
	419	return 0;
	420
	421	fail:
	422	while (--i >= 0)
	423	sysdev_remove_file(sysdev, cmm_attrs[i]);
	424	sysdev_unregister(sysdev);
	425	class_unregister:
	426	sysdev_class_unregister(&cmm_sysdev_class);
	427	return rc;
	428	}
	429
	430	/**
	431	* cmm_unregister_sysfs - Unregister from sysfs
	432	*
433	**/
434	static void cmm_unregister_sysfs(struct sys_device *sysdev)
435	{
436	int i;
437
438	for (i = 0; i < ARRAY_SIZE(cmm_attrs); i++)
439	sysdev_remove_file(sysdev, cmm_attrs[i]);
440	sysdev_unregister(sysdev);
441	sysdev_class_unregister(&cmm_sysdev_class);
442	}
443
fecba962 BK	444	/**
	445	* cmm_reboot_notifier - Make sure pages are not still marked as "loaned"
	446	*
	447	**/
	448	static int cmm_reboot_notifier(struct notifier_block *nb,
	449	unsigned long action, void *unused)
	450	{
	451	if (action == SYS_RESTART) {
	452	if (cmm_thread_ptr)
	453	kthread_stop(cmm_thread_ptr);
	454	cmm_thread_ptr = NULL;
	455	cmm_free_pages(loaned_pages);
	456	}
	457	return NOTIFY_DONE;
	458	}
	459
	460	static struct notifier_block cmm_reboot_nb = {
	461	.notifier_call = cmm_reboot_notifier,
	462	};
	463
14b8a76b RJ	464	/**
	465	* cmm_count_pages - Count the number of pages loaned in a particular range.
	466	*
	467	* @arg: memory_isolate_notify structure with address range and count
	468	*
	469	* Return value:
	470	* 0 on success
	471	**/
	472	static unsigned long cmm_count_pages(void *arg)
	473	{
	474	struct memory_isolate_notify *marg = arg;
	475	struct cmm_page_array *pa;
	476	unsigned long start = (unsigned long)pfn_to_kaddr(marg->start_pfn);
	477	unsigned long end = start + (marg->nr_pages << PAGE_SHIFT);
	478	unsigned long idx;
	479
	480	spin_lock(&cmm_lock);
	481	pa = cmm_page_list;
	482	while (pa) {
	483	if ((unsigned long)pa >= start && (unsigned long)pa < end)
	484	marg->pages_found++;
	485	for (idx = 0; idx < pa->index; idx++)
	486	if (pa->page[idx] >= start && pa->page[idx] < end)
	487	marg->pages_found++;
	488	pa = pa->next;
	489	}
	490	spin_unlock(&cmm_lock);
	491	return 0;
	492	}
	493
	494	/**
	495	* cmm_memory_isolate_cb - Handle memory isolation notifier calls
	496	* @self: notifier block struct
	497	* @action: action to take
	498	* @arg: struct memory_isolate_notify data for handler
	499	*
	500	* Return value:
	501	* NOTIFY_OK or notifier error based on subfunction return value
	502	**/
	503	static int cmm_memory_isolate_cb(struct notifier_block *self,
	504	unsigned long action, void *arg)
	505	{
	506	int ret = 0;
	507
	508	if (action == MEM_ISOLATE_COUNT)
	509	ret = cmm_count_pages(arg);
	510
	511	if (ret)
	512	ret = notifier_from_errno(ret);
	513	else
	514	ret = NOTIFY_OK;
	515
	516	return ret;
	517	}
	518
	519	static struct notifier_block cmm_mem_isolate_nb = {
	520	.notifier_call = cmm_memory_isolate_cb,
	521	.priority = CMM_MEM_ISOLATE_PRI
	522	};
	523
	524	/**
	525	* cmm_mem_going_offline - Unloan pages where memory is to be removed
	526	* @arg: memory_notify structure with page range to be offlined
	527	*
528	* Return value:
529	* 0 on success
530	**/
531	static int cmm_mem_going_offline(void *arg)
532	{
533	struct memory_notify *marg = arg;
534	unsigned long start_page = (unsigned long)pfn_to_kaddr(marg->start_pfn);
535	unsigned long end_page = start_page + (marg->nr_pages << PAGE_SHIFT);
536	struct cmm_page_array pa_curr, pa_last, *npa;
537	unsigned long idx;
538	unsigned long freed = 0;
539
540	cmm_dbg("Memory going offline, searching 0x%lx (%ld pages).\n",
541	start_page, marg->nr_pages);
542	spin_lock(&cmm_lock);
543
544	/* Search the page list for pages in the range to be offlined */
545	pa_last = pa_curr = cmm_page_list;
546	while (pa_curr) {
547	for (idx = (pa_curr->index - 1); (idx + 1) > 0; idx--) {
548	if ((pa_curr->page[idx] < start_page) \|\|
549	(pa_curr->page[idx] >= end_page))
550	continue;
551
552	plpar_page_set_active(__pa(pa_curr->page[idx]));
553	free_page(pa_curr->page[idx]);
554	freed++;
555	loaned_pages--;
556	totalram_pages++;
557	pa_curr->page[idx] = pa_last->page[--pa_last->index];
558	if (pa_last->index == 0) {
559	if (pa_curr == pa_last)
560	pa_curr = pa_last->next;
561	pa_last = pa_last->next;
562	free_page((unsigned long)cmm_page_list);
563	cmm_page_list = pa_last;
564	continue;
565	}
566	}
567	pa_curr = pa_curr->next;
568	}
569
570	/* Search for page list structures in the range to be offlined */
571	pa_last = NULL;
572	pa_curr = cmm_page_list;
573	while (pa_curr) {
574	if (((unsigned long)pa_curr >= start_page) &&
575	((unsigned long)pa_curr < end_page)) {
576	npa = (struct cmm_page_array *)__get_free_page(
577	GFP_NOIO \| __GFP_NOWARN \|
578	__GFP_NORETRY \| __GFP_NOMEMALLOC);
579	if (!npa) {
580	spin_unlock(&cmm_lock);
581	cmm_dbg("Failed to allocate memory for list "
582	"management. Memory hotplug "
583	"failed.\n");
584	return ENOMEM;
585	}
586	memcpy(npa, pa_curr, PAGE_SIZE);
587	if (pa_curr == cmm_page_list)
588	cmm_page_list = npa;
589	if (pa_last)
590	pa_last->next = npa;
591	free_page((unsigned long) pa_curr);
592	freed++;
593	pa_curr = npa;
594	}
595
596	pa_last = pa_curr;
597	pa_curr = pa_curr->next;
598	}
599
600	spin_unlock(&cmm_lock);
601	cmm_dbg("Released %ld pages in the search range.\n", freed);
602
603	return 0;
604	}
605
606	/**
607	* cmm_memory_cb - Handle memory hotplug notifier calls
608	* @self: notifier block struct
609	* @action: action to take
610	* @arg: struct memory_notify data for handler
611	*
612	* Return value:
613	* NOTIFY_OK or notifier error based on subfunction return value
614	*
615	**/
616	static int cmm_memory_cb(struct notifier_block *self,
617	unsigned long action, void *arg)
618	{
619	int ret = 0;
620
621	switch (action) {
622	case MEM_GOING_OFFLINE:
623	mutex_lock(&hotplug_mutex);
624	hotplug_occurred = 1;
625	ret = cmm_mem_going_offline(arg);
626	break;
627	case MEM_OFFLINE:
628	case MEM_CANCEL_OFFLINE:
629	mutex_unlock(&hotplug_mutex);
630	cmm_dbg("Memory offline operation complete.\n");
631	break;
632	case MEM_GOING_ONLINE:
633	case MEM_ONLINE:
634	case MEM_CANCEL_ONLINE:
635	break;
636	}
637
638	if (ret)
639	ret = notifier_from_errno(ret);
640	else
641	ret = NOTIFY_OK;
642
643	return ret;
644	}
645
646	static struct notifier_block cmm_mem_nb = {
647	.notifier_call = cmm_memory_cb,
648	.priority = CMM_MEM_HOTPLUG_PRI
649	};
650
84af458b BK	651	/**
	652	* cmm_init - Module initialization
	653	*
	654	* Return value:
	655	* 0 on success / other on failure
	656	**/
	657	static int cmm_init(void)
	658	{
	659	int rc = -ENOMEM;
	660
	661	if (!firmware_has_feature(FW_FEATURE_CMO))
	662	return -EOPNOTSUPP;
	663
	664	if ((rc = register_oom_notifier(&cmm_oom_nb)) < 0)
	665	return rc;
	666
fecba962	667	if ((rc = register_reboot_notifier(&cmm_reboot_nb)))
84af458b BK	668	goto out_oom_notifier;
84af458b BK	669
fecba962 BK	670	if ((rc = cmm_sysfs_register(&cmm_sysdev)))
	671	goto out_reboot_notifier;
	672
14b8a76b RJ	673	if (register_memory_notifier(&cmm_mem_nb) \|\|
	674	register_memory_isolate_notifier(&cmm_mem_isolate_nb))
	675	goto out_unregister_notifier;
	676
84af458b BK	677	if (cmm_disabled)
	678	return rc;
	679
	680	cmm_thread_ptr = kthread_run(cmm_thread, NULL, "cmmthread");
	681	if (IS_ERR(cmm_thread_ptr)) {
	682	rc = PTR_ERR(cmm_thread_ptr);
14b8a76b	683	goto out_unregister_notifier;
84af458b BK	684	}
	685
	686	return rc;
	687
14b8a76b RJ	688	out_unregister_notifier:
	689	unregister_memory_notifier(&cmm_mem_nb);
	690	unregister_memory_isolate_notifier(&cmm_mem_isolate_nb);
84af458b	691	cmm_unregister_sysfs(&cmm_sysdev);
fecba962 BK	692	out_reboot_notifier:
fecba962 BK	693	unregister_reboot_notifier(&cmm_reboot_nb);
84af458b BK	694	out_oom_notifier:
	695	unregister_oom_notifier(&cmm_oom_nb);
	696	return rc;
	697	}
	698
	699	/**
	700	* cmm_exit - Module exit
	701	*
	702	* Return value:
	703	* nothing
	704	**/
	705	static void cmm_exit(void)
	706	{
	707	if (cmm_thread_ptr)
	708	kthread_stop(cmm_thread_ptr);
	709	unregister_oom_notifier(&cmm_oom_nb);
fecba962	710	unregister_reboot_notifier(&cmm_reboot_nb);
14b8a76b RJ	711	unregister_memory_notifier(&cmm_mem_nb);
14b8a76b RJ	712	unregister_memory_isolate_notifier(&cmm_mem_isolate_nb);
84af458b BK	713	cmm_free_pages(loaned_pages);
	714	cmm_unregister_sysfs(&cmm_sysdev);
	715	}
	716
	717	/**
	718	* cmm_set_disable - Disable/Enable CMM
	719	*
	720	* Return value:
	721	* 0 on success / other on failure
	722	**/
	723	static int cmm_set_disable(const char val, struct kernel_param kp)
	724	{
	725	int disable = simple_strtoul(val, NULL, 10);
	726
	727	if (disable != 0 && disable != 1)
	728	return -EINVAL;
	729
	730	if (disable && !cmm_disabled) {
	731	if (cmm_thread_ptr)
	732	kthread_stop(cmm_thread_ptr);
	733	cmm_thread_ptr = NULL;
	734	cmm_free_pages(loaned_pages);
	735	} else if (!disable && cmm_disabled) {
	736	cmm_thread_ptr = kthread_run(cmm_thread, NULL, "cmmthread");
	737	if (IS_ERR(cmm_thread_ptr))
	738	return PTR_ERR(cmm_thread_ptr);
	739	}
	740
	741	cmm_disabled = disable;
	742	return 0;
	743	}
	744
	745	module_param_call(disable, cmm_set_disable, param_get_uint,
	746	&cmm_disabled, S_IRUGO \| S_IWUSR);
	747	MODULE_PARM_DESC(disable, "Disable CMM. Set to 1 to disable. "
	748	"[Default=" __stringify(CMM_DISABLE) "]");
	749
	750	module_init(cmm_init);
	751	module_exit(cmm_exit);