[linux.git] / mm / swap_slots.c

/*
 * Manage cache of swap slots to be used for and returned from
 * swap.
 *
 * Copyright(c) 2016 Intel Corporation.
 *
 * Author: Tim Chen <[email protected]>
 *
 * We allocate the swap slots from the global pool and put
 * it into local per cpu caches.  This has the advantage
 * of no needing to acquire the swap_info lock every time
 * we need a new slot.
 *
 * There is also opportunity to simply return the slot
 * to local caches without needing to acquire swap_info
 * lock.  We do not reuse the returned slots directly but
 * move them back to the global pool in a batch.  This
 * allows the slots to coaellesce and reduce fragmentation.
 *
 * The swap entry allocated is marked with SWAP_HAS_CACHE
 * flag in map_count that prevents it from being allocated
 * again from the global pool.
 *
 * The swap slots cache is protected by a mutex instead of
 * a spin lock as when we search for slots with scan_swap_map,
 * we can possibly sleep.
 */

#include <linux/swap_slots.h>
#include <linux/cpu.h>
#include <linux/cpumask.h>
#include <linux/vmalloc.h>
#include <linux/mutex.h>
#include <linux/mm.h>

#ifdef CONFIG_SWAP

static DEFINE_PER_CPU(struct swap_slots_cache, swp_slots);
static bool	swap_slot_cache_active;
bool	swap_slot_cache_enabled;
static bool	swap_slot_cache_initialized;
DEFINE_MUTEX(swap_slots_cache_mutex);
/* Serialize swap slots cache enable/disable operations */
DEFINE_MUTEX(swap_slots_cache_enable_mutex);

static void __drain_swap_slots_cache(unsigned int type);
static void deactivate_swap_slots_cache(void);
static void reactivate_swap_slots_cache(void);

#define use_swap_slot_cache (swap_slot_cache_active && \
		swap_slot_cache_enabled && swap_slot_cache_initialized)
#define SLOTS_CACHE 0x1
#define SLOTS_CACHE_RET 0x2

static void deactivate_swap_slots_cache(void)
{
	mutex_lock(&swap_slots_cache_mutex);
	swap_slot_cache_active = false;
	__drain_swap_slots_cache(SLOTS_CACHE|SLOTS_CACHE_RET);
	mutex_unlock(&swap_slots_cache_mutex);
}

static void reactivate_swap_slots_cache(void)
{
	mutex_lock(&swap_slots_cache_mutex);
	swap_slot_cache_active = true;
	mutex_unlock(&swap_slots_cache_mutex);
}

/* Must not be called with cpu hot plug lock */
void disable_swap_slots_cache_lock(void)
{
	mutex_lock(&swap_slots_cache_enable_mutex);
	swap_slot_cache_enabled = false;
	if (swap_slot_cache_initialized) {
		/* serialize with cpu hotplug operations */
		get_online_cpus();
		__drain_swap_slots_cache(SLOTS_CACHE|SLOTS_CACHE_RET);
		put_online_cpus();
	}
}

static void __reenable_swap_slots_cache(void)
{
	swap_slot_cache_enabled = has_usable_swap();
}

void reenable_swap_slots_cache_unlock(void)
{
	__reenable_swap_slots_cache();
	mutex_unlock(&swap_slots_cache_enable_mutex);
}

static bool check_cache_active(void)
{
	long pages;

	if (!swap_slot_cache_enabled || !swap_slot_cache_initialized)
		return false;

	pages = get_nr_swap_pages();
	if (!swap_slot_cache_active) {
		if (pages > num_online_cpus() *
		    THRESHOLD_ACTIVATE_SWAP_SLOTS_CACHE)
			reactivate_swap_slots_cache();
		goto out;
	}

	/* if global pool of slot caches too low, deactivate cache */
	if (pages < num_online_cpus() * THRESHOLD_DEACTIVATE_SWAP_SLOTS_CACHE)
		deactivate_swap_slots_cache();
out:
	return swap_slot_cache_active;
}

static int alloc_swap_slot_cache(unsigned int cpu)
{
	struct swap_slots_cache *cache;
	swp_entry_t *slots, *slots_ret;

	/*
	 * Do allocation outside swap_slots_cache_mutex
	 * as kvzalloc could trigger reclaim and get_swap_page,
	 * which can lock swap_slots_cache_mutex.
	 */
	slots = kvzalloc(sizeof(swp_entry_t) * SWAP_SLOTS_CACHE_SIZE,
			 GFP_KERNEL);
	if (!slots)
		return -ENOMEM;

	slots_ret = kvzalloc(sizeof(swp_entry_t) * SWAP_SLOTS_CACHE_SIZE,
			     GFP_KERNEL);
	if (!slots_ret) {
		kvfree(slots);
		return -ENOMEM;
	}

	mutex_lock(&swap_slots_cache_mutex);
	cache = &per_cpu(swp_slots, cpu);
	if (cache->slots || cache->slots_ret)
		/* cache already allocated */
		goto out;
	if (!cache->lock_initialized) {
		mutex_init(&cache->alloc_lock);
		spin_lock_init(&cache->free_lock);
		cache->lock_initialized = true;
	}
	cache->nr = 0;
	cache->cur = 0;
	cache->n_ret = 0;
	cache->slots = slots;
	slots = NULL;
	cache->slots_ret = slots_ret;
	slots_ret = NULL;
out:
	mutex_unlock(&swap_slots_cache_mutex);
	if (slots)
		kvfree(slots);
	if (slots_ret)
		kvfree(slots_ret);
	return 0;
}

static void drain_slots_cache_cpu(unsigned int cpu, unsigned int type,
				  bool free_slots)
{
	struct swap_slots_cache *cache;
	swp_entry_t *slots = NULL;

	cache = &per_cpu(swp_slots, cpu);
	if ((type & SLOTS_CACHE) && cache->slots) {
		mutex_lock(&cache->alloc_lock);
		swapcache_free_entries(cache->slots + cache->cur, cache->nr);
		cache->cur = 0;
		cache->nr = 0;
		if (free_slots && cache->slots) {
			kvfree(cache->slots);
			cache->slots = NULL;
		}
		mutex_unlock(&cache->alloc_lock);
	}
	if ((type & SLOTS_CACHE_RET) && cache->slots_ret) {
		spin_lock_irq(&cache->free_lock);
		swapcache_free_entries(cache->slots_ret, cache->n_ret);
		cache->n_ret = 0;
		if (free_slots && cache->slots_ret) {
			slots = cache->slots_ret;
			cache->slots_ret = NULL;
		}
		spin_unlock_irq(&cache->free_lock);
		if (slots)
			kvfree(slots);
	}
}

static void __drain_swap_slots_cache(unsigned int type)
{
	unsigned int cpu;

	/*
	 * This function is called during
	 *	1) swapoff, when we have to make sure no
	 *	   left over slots are in cache when we remove
	 *	   a swap device;
	 *      2) disabling of swap slot cache, when we run low
	 *	   on swap slots when allocating memory and need
	 *	   to return swap slots to global pool.
	 *
	 * We cannot acquire cpu hot plug lock here as
	 * this function can be invoked in the cpu
	 * hot plug path:
	 * cpu_up -> lock cpu_hotplug -> cpu hotplug state callback
	 *   -> memory allocation -> direct reclaim -> get_swap_page
	 *   -> drain_swap_slots_cache
	 *
	 * Hence the loop over current online cpu below could miss cpu that
	 * is being brought online but not yet marked as online.
	 * That is okay as we do not schedule and run anything on a
	 * cpu before it has been marked online. Hence, we will not
	 * fill any swap slots in slots cache of such cpu.
	 * There are no slots on such cpu that need to be drained.
	 */
	for_each_online_cpu(cpu)
		drain_slots_cache_cpu(cpu, type, false);
}

static int free_slot_cache(unsigned int cpu)
{
	mutex_lock(&swap_slots_cache_mutex);
	drain_slots_cache_cpu(cpu, SLOTS_CACHE | SLOTS_CACHE_RET, true);
	mutex_unlock(&swap_slots_cache_mutex);
	return 0;
}

int enable_swap_slots_cache(void)
{
	int ret = 0;

	mutex_lock(&swap_slots_cache_enable_mutex);
	if (swap_slot_cache_initialized) {
		__reenable_swap_slots_cache();
		goto out_unlock;
	}

	ret = cpuhp_setup_state(CPUHP_AP_ONLINE_DYN, "swap_slots_cache",
				alloc_swap_slot_cache, free_slot_cache);
	if (WARN_ONCE(ret < 0, "Cache allocation failed (%s), operating "
			       "without swap slots cache.\n", __func__))
		goto out_unlock;

	swap_slot_cache_initialized = true;
	__reenable_swap_slots_cache();
out_unlock:
	mutex_unlock(&swap_slots_cache_enable_mutex);
	return 0;
}

/* called with swap slot cache's alloc lock held */
static int refill_swap_slots_cache(struct swap_slots_cache *cache)
{
	if (!use_swap_slot_cache || cache->nr)
		return 0;

	cache->cur = 0;
	if (swap_slot_cache_active)
		cache->nr = get_swap_pages(SWAP_SLOTS_CACHE_SIZE, cache->slots);

	return cache->nr;
}

int free_swap_slot(swp_entry_t entry)
{
	struct swap_slots_cache *cache;

	cache = &get_cpu_var(swp_slots);
	if (use_swap_slot_cache && cache->slots_ret) {
		spin_lock_irq(&cache->free_lock);
		/* Swap slots cache may be deactivated before acquiring lock */
		if (!use_swap_slot_cache) {
			spin_unlock_irq(&cache->free_lock);
			goto direct_free;
		}
		if (cache->n_ret >= SWAP_SLOTS_CACHE_SIZE) {
			/*
			 * Return slots to global pool.
			 * The current swap_map value is SWAP_HAS_CACHE.
			 * Set it to 0 to indicate it is available for
			 * allocation in global pool
			 */
			swapcache_free_entries(cache->slots_ret, cache->n_ret);
			cache->n_ret = 0;
		}
		cache->slots_ret[cache->n_ret++] = entry;
		spin_unlock_irq(&cache->free_lock);
	} else {
direct_free:
		swapcache_free_entries(&entry, 1);
	}
	put_cpu_var(swp_slots);

	return 0;
}

swp_entry_t get_swap_page(void)
{
	swp_entry_t entry, *pentry;
	struct swap_slots_cache *cache;

	/*
	 * Preemption is allowed here, because we may sleep
	 * in refill_swap_slots_cache().  But it is safe, because
	 * accesses to the per-CPU data structure are protected by the
	 * mutex cache->alloc_lock.
	 *
	 * The alloc path here does not touch cache->slots_ret
	 * so cache->free_lock is not taken.
	 */
	cache = raw_cpu_ptr(&swp_slots);

	entry.val = 0;
	if (check_cache_active()) {
		mutex_lock(&cache->alloc_lock);
		if (cache->slots) {
repeat:
			if (cache->nr) {
				pentry = &cache->slots[cache->cur++];
				entry = *pentry;
				pentry->val = 0;
				cache->nr--;
			} else {
				if (refill_swap_slots_cache(cache))
					goto repeat;
			}
		}
		mutex_unlock(&cache->alloc_lock);
		if (entry.val)
			return entry;
	}

	get_swap_pages(1, &entry);

	return entry;
}

#endif /* CONFIG_SWAP */
Commit	Line	Data
67afa38e TC	1	/*
	2	* Manage cache of swap slots to be used for and returned from
	3	* swap.
	4	*
	5	* Copyright(c) 2016 Intel Corporation.
	6	*
	7	* Author: Tim Chen <[email protected]>
	8	*
	9	* We allocate the swap slots from the global pool and put
	10	* it into local per cpu caches. This has the advantage
	11	* of no needing to acquire the swap_info lock every time
	12	* we need a new slot.
	13	*
	14	* There is also opportunity to simply return the slot
	15	* to local caches without needing to acquire swap_info
	16	* lock. We do not reuse the returned slots directly but
	17	* move them back to the global pool in a batch. This
	18	* allows the slots to coaellesce and reduce fragmentation.
	19	*
	20	* The swap entry allocated is marked with SWAP_HAS_CACHE
	21	* flag in map_count that prevents it from being allocated
	22	* again from the global pool.
	23	*
	24	* The swap slots cache is protected by a mutex instead of
	25	* a spin lock as when we search for slots with scan_swap_map,
	26	* we can possibly sleep.
	27	*/
	28
	29	#include <linux/swap_slots.h>
	30	#include <linux/cpu.h>
	31	#include <linux/cpumask.h>
	32	#include <linux/vmalloc.h>
	33	#include <linux/mutex.h>
54f180d3	34	#include <linux/mm.h>
67afa38e TC	35
	36	#ifdef CONFIG_SWAP
	37
	38	static DEFINE_PER_CPU(struct swap_slots_cache, swp_slots);
	39	static bool swap_slot_cache_active;
ba81f838	40	bool swap_slot_cache_enabled;
67afa38e TC	41	static bool swap_slot_cache_initialized;
	42	DEFINE_MUTEX(swap_slots_cache_mutex);
	43	/* Serialize swap slots cache enable/disable operations */
	44	DEFINE_MUTEX(swap_slots_cache_enable_mutex);
	45
	46	static void __drain_swap_slots_cache(unsigned int type);
	47	static void deactivate_swap_slots_cache(void);
	48	static void reactivate_swap_slots_cache(void);
	49
	50	#define use_swap_slot_cache (swap_slot_cache_active && \
	51	swap_slot_cache_enabled && swap_slot_cache_initialized)
	52	#define SLOTS_CACHE 0x1
	53	#define SLOTS_CACHE_RET 0x2
	54
	55	static void deactivate_swap_slots_cache(void)
	56	{
	57	mutex_lock(&swap_slots_cache_mutex);
	58	swap_slot_cache_active = false;
	59	__drain_swap_slots_cache(SLOTS_CACHE\|SLOTS_CACHE_RET);
	60	mutex_unlock(&swap_slots_cache_mutex);
	61	}
	62
	63	static void reactivate_swap_slots_cache(void)
	64	{
	65	mutex_lock(&swap_slots_cache_mutex);
	66	swap_slot_cache_active = true;
	67	mutex_unlock(&swap_slots_cache_mutex);
	68	}
	69
	70	/* Must not be called with cpu hot plug lock */
	71	void disable_swap_slots_cache_lock(void)
	72	{
	73	mutex_lock(&swap_slots_cache_enable_mutex);
	74	swap_slot_cache_enabled = false;
	75	if (swap_slot_cache_initialized) {
	76	/* serialize with cpu hotplug operations */
	77	get_online_cpus();
	78	__drain_swap_slots_cache(SLOTS_CACHE\|SLOTS_CACHE_RET);
	79	put_online_cpus();
	80	}
	81	}
	82
	83	static void __reenable_swap_slots_cache(void)
	84	{
	85	swap_slot_cache_enabled = has_usable_swap();
	86	}
	87
	88	void reenable_swap_slots_cache_unlock(void)
	89	{
	90	__reenable_swap_slots_cache();
	91	mutex_unlock(&swap_slots_cache_enable_mutex);
	92	}
	93
	94	static bool check_cache_active(void)
	95	{
	96	long pages;
	97
	98	if (!swap_slot_cache_enabled \|\| !swap_slot_cache_initialized)
	99	return false;
	100
	101	pages = get_nr_swap_pages();
	102	if (!swap_slot_cache_active) {
	103	if (pages > num_online_cpus() *
	104	THRESHOLD_ACTIVATE_SWAP_SLOTS_CACHE)
105	reactivate_swap_slots_cache();
106	goto out;
107	}
108
109	/* if global pool of slot caches too low, deactivate cache */
110	if (pages < num_online_cpus() * THRESHOLD_DEACTIVATE_SWAP_SLOTS_CACHE)
111	deactivate_swap_slots_cache();
112	out:
113	return swap_slot_cache_active;
114	}
115
116	static int alloc_swap_slot_cache(unsigned int cpu)
117	{
118	struct swap_slots_cache *cache;
119	swp_entry_t slots, slots_ret;
120
121	/*
122	* Do allocation outside swap_slots_cache_mutex
54f180d3	123	* as kvzalloc could trigger reclaim and get_swap_page,
67afa38e TC	124	* which can lock swap_slots_cache_mutex.
67afa38e TC	125	*/
54f180d3 YH	126	slots = kvzalloc(sizeof(swp_entry_t) * SWAP_SLOTS_CACHE_SIZE,
54f180d3 YH	127	GFP_KERNEL);
67afa38e TC	128	if (!slots)
	129	return -ENOMEM;
	130
54f180d3 YH	131	slots_ret = kvzalloc(sizeof(swp_entry_t) * SWAP_SLOTS_CACHE_SIZE,
54f180d3 YH	132	GFP_KERNEL);
67afa38e	133	if (!slots_ret) {
54f180d3	134	kvfree(slots);
67afa38e TC	135	return -ENOMEM;
	136	}
	137
	138	mutex_lock(&swap_slots_cache_mutex);
	139	cache = &per_cpu(swp_slots, cpu);
	140	if (cache->slots \|\| cache->slots_ret)
	141	/* cache already allocated */
	142	goto out;
	143	if (!cache->lock_initialized) {
	144	mutex_init(&cache->alloc_lock);
	145	spin_lock_init(&cache->free_lock);
	146	cache->lock_initialized = true;
	147	}
	148	cache->nr = 0;
	149	cache->cur = 0;
	150	cache->n_ret = 0;
	151	cache->slots = slots;
	152	slots = NULL;
	153	cache->slots_ret = slots_ret;
	154	slots_ret = NULL;
	155	out:
	156	mutex_unlock(&swap_slots_cache_mutex);
	157	if (slots)
54f180d3	158	kvfree(slots);
67afa38e	159	if (slots_ret)
54f180d3	160	kvfree(slots_ret);
67afa38e TC	161	return 0;
	162	}
	163
	164	static void drain_slots_cache_cpu(unsigned int cpu, unsigned int type,
	165	bool free_slots)
	166	{
	167	struct swap_slots_cache *cache;
	168	swp_entry_t *slots = NULL;
	169
	170	cache = &per_cpu(swp_slots, cpu);
	171	if ((type & SLOTS_CACHE) && cache->slots) {
	172	mutex_lock(&cache->alloc_lock);
	173	swapcache_free_entries(cache->slots + cache->cur, cache->nr);
	174	cache->cur = 0;
	175	cache->nr = 0;
	176	if (free_slots && cache->slots) {
54f180d3	177	kvfree(cache->slots);
67afa38e TC	178	cache->slots = NULL;
	179	}
	180	mutex_unlock(&cache->alloc_lock);
	181	}
	182	if ((type & SLOTS_CACHE_RET) && cache->slots_ret) {
	183	spin_lock_irq(&cache->free_lock);
	184	swapcache_free_entries(cache->slots_ret, cache->n_ret);
	185	cache->n_ret = 0;
	186	if (free_slots && cache->slots_ret) {
	187	slots = cache->slots_ret;
	188	cache->slots_ret = NULL;
	189	}
	190	spin_unlock_irq(&cache->free_lock);
	191	if (slots)
54f180d3	192	kvfree(slots);
67afa38e TC	193	}
	194	}
	195
	196	static void __drain_swap_slots_cache(unsigned int type)
	197	{
	198	unsigned int cpu;
	199
	200	/*
	201	* This function is called during
	202	* 1) swapoff, when we have to make sure no
	203	* left over slots are in cache when we remove
	204	* a swap device;
	205	* 2) disabling of swap slot cache, when we run low
	206	* on swap slots when allocating memory and need
	207	* to return swap slots to global pool.
	208	*
	209	* We cannot acquire cpu hot plug lock here as
	210	* this function can be invoked in the cpu
	211	* hot plug path:
	212	* cpu_up -> lock cpu_hotplug -> cpu hotplug state callback
	213	* -> memory allocation -> direct reclaim -> get_swap_page
	214	* -> drain_swap_slots_cache
	215	*
	216	* Hence the loop over current online cpu below could miss cpu that
	217	* is being brought online but not yet marked as online.
	218	* That is okay as we do not schedule and run anything on a
	219	* cpu before it has been marked online. Hence, we will not
	220	* fill any swap slots in slots cache of such cpu.
	221	* There are no slots on such cpu that need to be drained.
	222	*/
	223	for_each_online_cpu(cpu)
	224	drain_slots_cache_cpu(cpu, type, false);
	225	}
	226
	227	static int free_slot_cache(unsigned int cpu)
	228	{
	229	mutex_lock(&swap_slots_cache_mutex);
	230	drain_slots_cache_cpu(cpu, SLOTS_CACHE \| SLOTS_CACHE_RET, true);
	231	mutex_unlock(&swap_slots_cache_mutex);
	232	return 0;
	233	}
	234
	235	int enable_swap_slots_cache(void)
	236	{
	237	int ret = 0;
	238
	239	mutex_lock(&swap_slots_cache_enable_mutex);
	240	if (swap_slot_cache_initialized) {
	241	__reenable_swap_slots_cache();
	242	goto out_unlock;
	243	}
	244
	245	ret = cpuhp_setup_state(CPUHP_AP_ONLINE_DYN, "swap_slots_cache",
	246	alloc_swap_slot_cache, free_slot_cache);
9b7a8143 TC	247	if (WARN_ONCE(ret < 0, "Cache allocation failed (%s), operating "
9b7a8143 TC	248	"without swap slots cache.\n", __func__))
67afa38e	249	goto out_unlock;
9b7a8143	250
67afa38e TC	251	swap_slot_cache_initialized = true;
	252	__reenable_swap_slots_cache();
	253	out_unlock:
	254	mutex_unlock(&swap_slots_cache_enable_mutex);
	255	return 0;
	256	}
	257
	258	/* called with swap slot cache's alloc lock held */
	259	static int refill_swap_slots_cache(struct swap_slots_cache *cache)
	260	{
	261	if (!use_swap_slot_cache \|\| cache->nr)
	262	return 0;
	263
	264	cache->cur = 0;
	265	if (swap_slot_cache_active)
	266	cache->nr = get_swap_pages(SWAP_SLOTS_CACHE_SIZE, cache->slots);
	267
	268	return cache->nr;
	269	}
	270
	271	int free_swap_slot(swp_entry_t entry)
	272	{
	273	struct swap_slots_cache *cache;
	274
67afa38e TC	275	cache = &get_cpu_var(swp_slots);
	276	if (use_swap_slot_cache && cache->slots_ret) {
	277	spin_lock_irq(&cache->free_lock);
	278	/* Swap slots cache may be deactivated before acquiring lock */
	279	if (!use_swap_slot_cache) {
	280	spin_unlock_irq(&cache->free_lock);
	281	goto direct_free;
	282	}
	283	if (cache->n_ret >= SWAP_SLOTS_CACHE_SIZE) {
	284	/*
	285	* Return slots to global pool.
	286	* The current swap_map value is SWAP_HAS_CACHE.
	287	* Set it to 0 to indicate it is available for
	288	* allocation in global pool
	289	*/
	290	swapcache_free_entries(cache->slots_ret, cache->n_ret);
	291	cache->n_ret = 0;
	292	}
	293	cache->slots_ret[cache->n_ret++] = entry;
	294	spin_unlock_irq(&cache->free_lock);
	295	} else {
	296	direct_free:
	297	swapcache_free_entries(&entry, 1);
	298	}
	299	put_cpu_var(swp_slots);
	300
	301	return 0;
	302	}
	303
	304	swp_entry_t get_swap_page(void)
	305	{
	306	swp_entry_t entry, *pentry;
	307	struct swap_slots_cache *cache;
	308
	309	/*
	310	* Preemption is allowed here, because we may sleep
	311	* in refill_swap_slots_cache(). But it is safe, because
	312	* accesses to the per-CPU data structure are protected by the
	313	* mutex cache->alloc_lock.
	314	*
	315	* The alloc path here does not touch cache->slots_ret
	316	* so cache->free_lock is not taken.
	317	*/
	318	cache = raw_cpu_ptr(&swp_slots);
	319
	320	entry.val = 0;
	321	if (check_cache_active()) {
	322	mutex_lock(&cache->alloc_lock);
	323	if (cache->slots) {
	324	repeat:
	325	if (cache->nr) {
	326	pentry = &cache->slots[cache->cur++];
	327	entry = *pentry;
	328	pentry->val = 0;
	329	cache->nr--;
	330	} else {
	331	if (refill_swap_slots_cache(cache))
	332	goto repeat;
	333	}
	334	}
	335	mutex_unlock(&cache->alloc_lock);
	336	if (entry.val)
	337	return entry;
	338	}
339
340	get_swap_pages(1, &entry);
341
342	return entry;
343	}
344
345	#endif /* CONFIG_SWAP */