[linux.git] / mm / memremap.c

/* SPDX-License-Identifier: GPL-2.0 */
/* Copyright(c) 2015 Intel Corporation. All rights reserved. */
#include <linux/device.h>
#include <linux/io.h>
#include <linux/kasan.h>
#include <linux/memory_hotplug.h>
#include <linux/mm.h>
#include <linux/pfn_t.h>
#include <linux/swap.h>
#include <linux/swapops.h>
#include <linux/types.h>
#include <linux/wait_bit.h>
#include <linux/xarray.h>

static DEFINE_XARRAY(pgmap_array);

#ifdef CONFIG_DEV_PAGEMAP_OPS
DEFINE_STATIC_KEY_FALSE(devmap_managed_key);
EXPORT_SYMBOL(devmap_managed_key);
static atomic_t devmap_managed_enable;

static void devmap_managed_enable_put(void)
{
	if (atomic_dec_and_test(&devmap_managed_enable))
		static_branch_disable(&devmap_managed_key);
}

static int devmap_managed_enable_get(struct dev_pagemap *pgmap)
{
	if (pgmap->type == MEMORY_DEVICE_PRIVATE &&
	    (!pgmap->ops || !pgmap->ops->page_free)) {
		WARN(1, "Missing page_free method\n");
		return -EINVAL;
	}

	if (atomic_inc_return(&devmap_managed_enable) == 1)
		static_branch_enable(&devmap_managed_key);
	return 0;
}
#else
static int devmap_managed_enable_get(struct dev_pagemap *pgmap)
{
	return -EINVAL;
}
static void devmap_managed_enable_put(void)
{
}
#endif /* CONFIG_DEV_PAGEMAP_OPS */

static void pgmap_array_delete(struct resource *res)
{
	xa_store_range(&pgmap_array, PHYS_PFN(res->start), PHYS_PFN(res->end),
			NULL, GFP_KERNEL);
	synchronize_rcu();
}

static unsigned long pfn_first(struct dev_pagemap *pgmap)
{
	return PHYS_PFN(pgmap->res.start) +
		vmem_altmap_offset(pgmap_altmap(pgmap));
}

static unsigned long pfn_end(struct dev_pagemap *pgmap)
{
	const struct resource *res = &pgmap->res;

	return (res->start + resource_size(res)) >> PAGE_SHIFT;
}

static unsigned long pfn_next(unsigned long pfn)
{
	if (pfn % 1024 == 0)
		cond_resched();
	return pfn + 1;
}

#define for_each_device_pfn(pfn, map) \
	for (pfn = pfn_first(map); pfn < pfn_end(map); pfn = pfn_next(pfn))

static void dev_pagemap_kill(struct dev_pagemap *pgmap)
{
	if (pgmap->ops && pgmap->ops->kill)
		pgmap->ops->kill(pgmap);
	else
		percpu_ref_kill(pgmap->ref);
}

static void dev_pagemap_cleanup(struct dev_pagemap *pgmap)
{
	if (pgmap->ops && pgmap->ops->cleanup) {
		pgmap->ops->cleanup(pgmap);
	} else {
		wait_for_completion(&pgmap->done);
		percpu_ref_exit(pgmap->ref);
	}
	/*
	 * Undo the pgmap ref assignment for the internal case as the
	 * caller may re-enable the same pgmap.
	 */
	if (pgmap->ref == &pgmap->internal_ref)
		pgmap->ref = NULL;
}

void memunmap_pages(struct dev_pagemap *pgmap)
{
	struct resource *res = &pgmap->res;
	struct page *first_page;
	unsigned long pfn;
	int nid;

	dev_pagemap_kill(pgmap);
	for_each_device_pfn(pfn, pgmap)
		put_page(pfn_to_page(pfn));
	dev_pagemap_cleanup(pgmap);

	/* make sure to access a memmap that was actually initialized */
	first_page = pfn_to_page(pfn_first(pgmap));

	/* pages are dead and unused, undo the arch mapping */
	nid = page_to_nid(first_page);

	mem_hotplug_begin();
	remove_pfn_range_from_zone(page_zone(first_page), PHYS_PFN(res->start),
				   PHYS_PFN(resource_size(res)));
	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
		__remove_pages(PHYS_PFN(res->start),
			       PHYS_PFN(resource_size(res)), NULL);
	} else {
		arch_remove_memory(nid, res->start, resource_size(res),
				pgmap_altmap(pgmap));
		kasan_remove_zero_shadow(__va(res->start), resource_size(res));
	}
	mem_hotplug_done();

	untrack_pfn(NULL, PHYS_PFN(res->start), resource_size(res));
	pgmap_array_delete(res);
	WARN_ONCE(pgmap->altmap.alloc, "failed to free all reserved pages\n");
	devmap_managed_enable_put();
}
EXPORT_SYMBOL_GPL(memunmap_pages);

static void devm_memremap_pages_release(void *data)
{
	memunmap_pages(data);
}

static void dev_pagemap_percpu_release(struct percpu_ref *ref)
{
	struct dev_pagemap *pgmap =
		container_of(ref, struct dev_pagemap, internal_ref);

	complete(&pgmap->done);
}

/*
 * Not device managed version of dev_memremap_pages, undone by
 * memunmap_pages().  Please use dev_memremap_pages if you have a struct
 * device available.
 */
void *memremap_pages(struct dev_pagemap *pgmap, int nid)
{
	struct resource *res = &pgmap->res;
	struct dev_pagemap *conflict_pgmap;
	struct mhp_restrictions restrictions = {
		/*
		 * We do not want any optional features only our own memmap
		 */
		.altmap = pgmap_altmap(pgmap),
	};
	pgprot_t pgprot = PAGE_KERNEL;
	int error, is_ram;
	bool need_devmap_managed = true;

	switch (pgmap->type) {
	case MEMORY_DEVICE_PRIVATE:
		if (!IS_ENABLED(CONFIG_DEVICE_PRIVATE)) {
			WARN(1, "Device private memory not supported\n");
			return ERR_PTR(-EINVAL);
		}
		if (!pgmap->ops || !pgmap->ops->migrate_to_ram) {
			WARN(1, "Missing migrate_to_ram method\n");
			return ERR_PTR(-EINVAL);
		}
		if (!pgmap->owner) {
			WARN(1, "Missing owner\n");
			return ERR_PTR(-EINVAL);
		}
		break;
	case MEMORY_DEVICE_FS_DAX:
		if (!IS_ENABLED(CONFIG_ZONE_DEVICE) ||
		    IS_ENABLED(CONFIG_FS_DAX_LIMITED)) {
			WARN(1, "File system DAX not supported\n");
			return ERR_PTR(-EINVAL);
		}
		break;
	case MEMORY_DEVICE_DEVDAX:
	case MEMORY_DEVICE_PCI_P2PDMA:
		need_devmap_managed = false;
		break;
	default:
		WARN(1, "Invalid pgmap type %d\n", pgmap->type);
		break;
	}

	if (!pgmap->ref) {
		if (pgmap->ops && (pgmap->ops->kill || pgmap->ops->cleanup))
			return ERR_PTR(-EINVAL);

		init_completion(&pgmap->done);
		error = percpu_ref_init(&pgmap->internal_ref,
				dev_pagemap_percpu_release, 0, GFP_KERNEL);
		if (error)
			return ERR_PTR(error);
		pgmap->ref = &pgmap->internal_ref;
	} else {
		if (!pgmap->ops || !pgmap->ops->kill || !pgmap->ops->cleanup) {
			WARN(1, "Missing reference count teardown definition\n");
			return ERR_PTR(-EINVAL);
		}
	}

	if (need_devmap_managed) {
		error = devmap_managed_enable_get(pgmap);
		if (error)
			return ERR_PTR(error);
	}

	conflict_pgmap = get_dev_pagemap(PHYS_PFN(res->start), NULL);
	if (conflict_pgmap) {
		WARN(1, "Conflicting mapping in same section\n");
		put_dev_pagemap(conflict_pgmap);
		error = -ENOMEM;
		goto err_array;
	}

	conflict_pgmap = get_dev_pagemap(PHYS_PFN(res->end), NULL);
	if (conflict_pgmap) {
		WARN(1, "Conflicting mapping in same section\n");
		put_dev_pagemap(conflict_pgmap);
		error = -ENOMEM;
		goto err_array;
	}

	is_ram = region_intersects(res->start, resource_size(res),
		IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE);

	if (is_ram != REGION_DISJOINT) {
		WARN_ONCE(1, "%s attempted on %s region %pr\n", __func__,
				is_ram == REGION_MIXED ? "mixed" : "ram", res);
		error = -ENXIO;
		goto err_array;
	}

	error = xa_err(xa_store_range(&pgmap_array, PHYS_PFN(res->start),
				PHYS_PFN(res->end), pgmap, GFP_KERNEL));
	if (error)
		goto err_array;

	if (nid < 0)
		nid = numa_mem_id();

	error = track_pfn_remap(NULL, &pgprot, PHYS_PFN(res->start), 0,
			resource_size(res));
	if (error)
		goto err_pfn_remap;

	mem_hotplug_begin();

	/*
	 * For device private memory we call add_pages() as we only need to
	 * allocate and initialize struct page for the device memory. More-
	 * over the device memory is un-accessible thus we do not want to
	 * create a linear mapping for the memory like arch_add_memory()
	 * would do.
	 *
	 * For all other device memory types, which are accessible by
	 * the CPU, we do want the linear mapping and thus use
	 * arch_add_memory().
	 */
	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
		error = add_pages(nid, PHYS_PFN(res->start),
				PHYS_PFN(resource_size(res)), &restrictions);
	} else {
		error = kasan_add_zero_shadow(__va(res->start), resource_size(res));
		if (error) {
			mem_hotplug_done();
			goto err_kasan;
		}

		error = arch_add_memory(nid, res->start, resource_size(res),
					&restrictions);
	}

	if (!error) {
		struct zone *zone;

		zone = &NODE_DATA(nid)->node_zones[ZONE_DEVICE];
		move_pfn_range_to_zone(zone, PHYS_PFN(res->start),
				PHYS_PFN(resource_size(res)), restrictions.altmap);
	}

	mem_hotplug_done();
	if (error)
		goto err_add_memory;

	/*
	 * Initialization of the pages has been deferred until now in order
	 * to allow us to do the work while not holding the hotplug lock.
	 */
	memmap_init_zone_device(&NODE_DATA(nid)->node_zones[ZONE_DEVICE],
				PHYS_PFN(res->start),
				PHYS_PFN(resource_size(res)), pgmap);
	percpu_ref_get_many(pgmap->ref, pfn_end(pgmap) - pfn_first(pgmap));
	return __va(res->start);

 err_add_memory:
	kasan_remove_zero_shadow(__va(res->start), resource_size(res));
 err_kasan:
	untrack_pfn(NULL, PHYS_PFN(res->start), resource_size(res));
 err_pfn_remap:
	pgmap_array_delete(res);
 err_array:
	dev_pagemap_kill(pgmap);
	dev_pagemap_cleanup(pgmap);
	devmap_managed_enable_put();
	return ERR_PTR(error);
}
EXPORT_SYMBOL_GPL(memremap_pages);

/**
 * devm_memremap_pages - remap and provide memmap backing for the given resource
 * @dev: hosting device for @res
 * @pgmap: pointer to a struct dev_pagemap
 *
 * Notes:
 * 1/ At a minimum the res and type members of @pgmap must be initialized
 *    by the caller before passing it to this function
 *
 * 2/ The altmap field may optionally be initialized, in which case
 *    PGMAP_ALTMAP_VALID must be set in pgmap->flags.
 *
 * 3/ The ref field may optionally be provided, in which pgmap->ref must be
 *    'live' on entry and will be killed and reaped at
 *    devm_memremap_pages_release() time, or if this routine fails.
 *
 * 4/ res is expected to be a host memory range that could feasibly be
 *    treated as a "System RAM" range, i.e. not a device mmio range, but
 *    this is not enforced.
 */
void *devm_memremap_pages(struct device *dev, struct dev_pagemap *pgmap)
{
	int error;
	void *ret;

	ret = memremap_pages(pgmap, dev_to_node(dev));
	if (IS_ERR(ret))
		return ret;

	error = devm_add_action_or_reset(dev, devm_memremap_pages_release,
			pgmap);
	if (error)
		return ERR_PTR(error);
	return ret;
}
EXPORT_SYMBOL_GPL(devm_memremap_pages);

void devm_memunmap_pages(struct device *dev, struct dev_pagemap *pgmap)
{
	devm_release_action(dev, devm_memremap_pages_release, pgmap);
}
EXPORT_SYMBOL_GPL(devm_memunmap_pages);

unsigned long vmem_altmap_offset(struct vmem_altmap *altmap)
{
	/* number of pfns from base where pfn_to_page() is valid */
	if (altmap)
		return altmap->reserve + altmap->free;
	return 0;
}

void vmem_altmap_free(struct vmem_altmap *altmap, unsigned long nr_pfns)
{
	altmap->alloc -= nr_pfns;
}

/**
 * get_dev_pagemap() - take a new live reference on the dev_pagemap for @pfn
 * @pfn: page frame number to lookup page_map
 * @pgmap: optional known pgmap that already has a reference
 *
 * If @pgmap is non-NULL and covers @pfn it will be returned as-is.  If @pgmap
 * is non-NULL but does not cover @pfn the reference to it will be released.
 */
struct dev_pagemap *get_dev_pagemap(unsigned long pfn,
		struct dev_pagemap *pgmap)
{
	resource_size_t phys = PFN_PHYS(pfn);

	/*
	 * In the cached case we're already holding a live reference.
	 */
	if (pgmap) {
		if (phys >= pgmap->res.start && phys <= pgmap->res.end)
			return pgmap;
		put_dev_pagemap(pgmap);
	}

	/* fall back to slow path lookup */
	rcu_read_lock();
	pgmap = xa_load(&pgmap_array, PHYS_PFN(phys));
	if (pgmap && !percpu_ref_tryget_live(pgmap->ref))
		pgmap = NULL;
	rcu_read_unlock();

	return pgmap;
}
EXPORT_SYMBOL_GPL(get_dev_pagemap);

#ifdef CONFIG_DEV_PAGEMAP_OPS
void free_devmap_managed_page(struct page *page)
{
	/* notify page idle for dax */
	if (!is_device_private_page(page)) {
		wake_up_var(&page->_refcount);
		return;
	}

	/* Clear Active bit in case of parallel mark_page_accessed */
	__ClearPageActive(page);
	__ClearPageWaiters(page);

	mem_cgroup_uncharge(page);

	/*
	 * When a device_private page is freed, the page->mapping field
	 * may still contain a (stale) mapping value. For example, the
	 * lower bits of page->mapping may still identify the page as an
	 * anonymous page. Ultimately, this entire field is just stale
	 * and wrong, and it will cause errors if not cleared.  One
	 * example is:
	 *
	 *  migrate_vma_pages()
	 *    migrate_vma_insert_page()
	 *      page_add_new_anon_rmap()
	 *        __page_set_anon_rmap()
	 *          ...checks page->mapping, via PageAnon(page) call,
	 *            and incorrectly concludes that the page is an
	 *            anonymous page. Therefore, it incorrectly,
	 *            silently fails to set up the new anon rmap.
	 *
	 * For other types of ZONE_DEVICE pages, migration is either
	 * handled differently or not done at all, so there is no need
	 * to clear page->mapping.
	 */
	page->mapping = NULL;
	page->pgmap->ops->page_free(page);
}
#endif /* CONFIG_DEV_PAGEMAP_OPS */
Commit	Line	Data
5981690d DW	1	/* SPDX-License-Identifier: GPL-2.0 */
5981690d DW	2	/* Copyright(c) 2015 Intel Corporation. All rights reserved. */
7d3dcf26	3	#include <linux/device.h>
92281dee	4	#include <linux/io.h>
0207df4f	5	#include <linux/kasan.h>
41e94a85	6	#include <linux/memory_hotplug.h>
bcfa4b72 MW	7	#include <linux/mm.h>
bcfa4b72 MW	8	#include <linux/pfn_t.h>
5042db43 JG	9	#include <linux/swap.h>
5042db43 JG	10	#include <linux/swapops.h>
bcfa4b72	11	#include <linux/types.h>
e7638488	12	#include <linux/wait_bit.h>
bcfa4b72	13	#include <linux/xarray.h>
92281dee	14
bcfa4b72	15	static DEFINE_XARRAY(pgmap_array);
9476df7d	16
f6a55e1a CH	17	#ifdef CONFIG_DEV_PAGEMAP_OPS
	18	DEFINE_STATIC_KEY_FALSE(devmap_managed_key);
	19	EXPORT_SYMBOL(devmap_managed_key);
	20	static atomic_t devmap_managed_enable;
	21
6f42193f	22	static void devmap_managed_enable_put(void)
f6a55e1a CH	23	{
	24	if (atomic_dec_and_test(&devmap_managed_enable))
	25	static_branch_disable(&devmap_managed_key);
	26	}
	27
6f42193f	28	static int devmap_managed_enable_get(struct dev_pagemap *pgmap)
f6a55e1a	29	{
429589d6 DW	30	if (pgmap->type == MEMORY_DEVICE_PRIVATE &&
429589d6 DW	31	(!pgmap->ops \|\| !pgmap->ops->page_free)) {
f6a55e1a CH	32	WARN(1, "Missing page_free method\n");
	33	return -EINVAL;
	34	}
	35
	36	if (atomic_inc_return(&devmap_managed_enable) == 1)
	37	static_branch_enable(&devmap_managed_key);
6f42193f	38	return 0;
f6a55e1a CH	39	}
f6a55e1a CH	40	#else
6f42193f	41	static int devmap_managed_enable_get(struct dev_pagemap *pgmap)
f6a55e1a CH	42	{
	43	return -EINVAL;
	44	}
6f42193f CH	45	static void devmap_managed_enable_put(void)
	46	{
	47	}
f6a55e1a CH	48	#endif /* CONFIG_DEV_PAGEMAP_OPS */
f6a55e1a CH	49
bcfa4b72	50	static void pgmap_array_delete(struct resource *res)
ab1b597e	51	{
bcfa4b72 MW	52	xa_store_range(&pgmap_array, PHYS_PFN(res->start), PHYS_PFN(res->end),
bcfa4b72 MW	53	NULL, GFP_KERNEL);
ab1b597e	54	synchronize_rcu();
9476df7d DW	55	}
9476df7d DW	56
e7744aa2	57	static unsigned long pfn_first(struct dev_pagemap *pgmap)
5c2c2587	58	{
7cc7867f	59	return PHYS_PFN(pgmap->res.start) +
514caf23	60	vmem_altmap_offset(pgmap_altmap(pgmap));
5c2c2587 DW	61	}
5c2c2587 DW	62
e7744aa2	63	static unsigned long pfn_end(struct dev_pagemap *pgmap)
5c2c2587	64	{
e7744aa2	65	const struct resource *res = &pgmap->res;
5c2c2587 DW	66
	67	return (res->start + resource_size(res)) >> PAGE_SHIFT;
	68	}
	69
949b9325 DW	70	static unsigned long pfn_next(unsigned long pfn)
	71	{
	72	if (pfn % 1024 == 0)
	73	cond_resched();
	74	return pfn + 1;
	75	}
	76
5c2c2587	77	#define for_each_device_pfn(pfn, map) \
949b9325	78	for (pfn = pfn_first(map); pfn < pfn_end(map); pfn = pfn_next(pfn))
5c2c2587	79
24917f6b CH	80	static void dev_pagemap_kill(struct dev_pagemap *pgmap)
	81	{
	82	if (pgmap->ops && pgmap->ops->kill)
	83	pgmap->ops->kill(pgmap);
	84	else
	85	percpu_ref_kill(pgmap->ref);
	86	}
	87
	88	static void dev_pagemap_cleanup(struct dev_pagemap *pgmap)
	89	{
	90	if (pgmap->ops && pgmap->ops->cleanup) {
	91	pgmap->ops->cleanup(pgmap);
	92	} else {
	93	wait_for_completion(&pgmap->done);
	94	percpu_ref_exit(pgmap->ref);
	95	}
06282373 DW	96	/*
	97	* Undo the pgmap ref assignment for the internal case as the
	98	* caller may re-enable the same pgmap.
	99	*/
	100	if (pgmap->ref == &pgmap->internal_ref)
	101	pgmap->ref = NULL;
24917f6b CH	102	}
24917f6b CH	103
6869b7b2	104	void memunmap_pages(struct dev_pagemap *pgmap)
41e94a85	105	{
e7744aa2	106	struct resource *res = &pgmap->res;
77e080e7	107	struct page *first_page;
71389703	108	unsigned long pfn;
2c2a5af6	109	int nid;
71389703	110
24917f6b	111	dev_pagemap_kill(pgmap);
e7744aa2	112	for_each_device_pfn(pfn, pgmap)
71389703	113	put_page(pfn_to_page(pfn));
24917f6b	114	dev_pagemap_cleanup(pgmap);
9476df7d	115
77e080e7 AK	116	/* make sure to access a memmap that was actually initialized */
	117	first_page = pfn_to_page(pfn_first(pgmap));
	118
41e94a85	119	/* pages are dead and unused, undo the arch mapping */
77e080e7	120	nid = page_to_nid(first_page);
2c2a5af6	121
f931ab47	122	mem_hotplug_begin();
d33695b1 DH	123	remove_pfn_range_from_zone(page_zone(first_page), PHYS_PFN(res->start),
d33695b1 DH	124	PHYS_PFN(resource_size(res)));
69324b8f	125	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
feee6b29	126	__remove_pages(PHYS_PFN(res->start),
77e080e7	127	PHYS_PFN(resource_size(res)), NULL);
69324b8f	128	} else {
7cc7867f	129	arch_remove_memory(nid, res->start, resource_size(res),
514caf23	130	pgmap_altmap(pgmap));
7cc7867f	131	kasan_remove_zero_shadow(__va(res->start), resource_size(res));
69324b8f	132	}
f931ab47	133	mem_hotplug_done();
b5d24fda	134
7cc7867f	135	untrack_pfn(NULL, PHYS_PFN(res->start), resource_size(res));
bcfa4b72	136	pgmap_array_delete(res);
fdc029b1	137	WARN_ONCE(pgmap->altmap.alloc, "failed to free all reserved pages\n");
6f42193f	138	devmap_managed_enable_put();
9476df7d	139	}
6869b7b2 CH	140	EXPORT_SYMBOL_GPL(memunmap_pages);
	141
	142	static void devm_memremap_pages_release(void *data)
	143	{
	144	memunmap_pages(data);
	145	}
9476df7d	146
24917f6b CH	147	static void dev_pagemap_percpu_release(struct percpu_ref *ref)
	148	{
	149	struct dev_pagemap *pgmap =
	150	container_of(ref, struct dev_pagemap, internal_ref);
	151
	152	complete(&pgmap->done);
	153	}
	154
6869b7b2 CH	155	/*
	156	* Not device managed version of dev_memremap_pages, undone by
	157	* memunmap_pages(). Please use dev_memremap_pages if you have a struct
	158	* device available.
4b94ffdc	159	*/
6869b7b2	160	void memremap_pages(struct dev_pagemap pgmap, int nid)
41e94a85	161	{
949b9325	162	struct resource *res = &pgmap->res;
966cf44f	163	struct dev_pagemap *conflict_pgmap;
940519f0 MH	164	struct mhp_restrictions restrictions = {
	165	/*
	166	* We do not want any optional features only our own memmap
7cc7867f	167	*/
514caf23	168	.altmap = pgmap_altmap(pgmap),
940519f0	169	};
9049771f	170	pgprot_t pgprot = PAGE_KERNEL;
6869b7b2	171	int error, is_ram;
f6a55e1a	172	bool need_devmap_managed = true;
5f29a77c	173
3ed2dcdf CH	174	switch (pgmap->type) {
	175	case MEMORY_DEVICE_PRIVATE:
	176	if (!IS_ENABLED(CONFIG_DEVICE_PRIVATE)) {
	177	WARN(1, "Device private memory not supported\n");
	178	return ERR_PTR(-EINVAL);
	179	}
897e6365 CH	180	if (!pgmap->ops \|\| !pgmap->ops->migrate_to_ram) {
	181	WARN(1, "Missing migrate_to_ram method\n");
	182	return ERR_PTR(-EINVAL);
	183	}
f894ddd5 CH	184	if (!pgmap->owner) {
	185	WARN(1, "Missing owner\n");
	186	return ERR_PTR(-EINVAL);
	187	}
3ed2dcdf CH	188	break;
	189	case MEMORY_DEVICE_FS_DAX:
	190	if (!IS_ENABLED(CONFIG_ZONE_DEVICE) \|\|
	191	IS_ENABLED(CONFIG_FS_DAX_LIMITED)) {
	192	WARN(1, "File system DAX not supported\n");
	193	return ERR_PTR(-EINVAL);
	194	}
	195	break;
	196	case MEMORY_DEVICE_DEVDAX:
	197	case MEMORY_DEVICE_PCI_P2PDMA:
f6a55e1a	198	need_devmap_managed = false;
3ed2dcdf CH	199	break;
	200	default:
	201	WARN(1, "Invalid pgmap type %d\n", pgmap->type);
	202	break;
	203	}
	204
24917f6b CH	205	if (!pgmap->ref) {
	206	if (pgmap->ops && (pgmap->ops->kill \|\| pgmap->ops->cleanup))
	207	return ERR_PTR(-EINVAL);
	208
	209	init_completion(&pgmap->done);
	210	error = percpu_ref_init(&pgmap->internal_ref,
	211	dev_pagemap_percpu_release, 0, GFP_KERNEL);
	212	if (error)
	213	return ERR_PTR(error);
	214	pgmap->ref = &pgmap->internal_ref;
	215	} else {
	216	if (!pgmap->ops \|\| !pgmap->ops->kill \|\| !pgmap->ops->cleanup) {
	217	WARN(1, "Missing reference count teardown definition\n");
	218	return ERR_PTR(-EINVAL);
	219	}
50f44ee7	220	}
a95c90f1	221
f6a55e1a	222	if (need_devmap_managed) {
6f42193f	223	error = devmap_managed_enable_get(pgmap);
f6a55e1a CH	224	if (error)
	225	return ERR_PTR(error);
	226	}
	227
7cc7867f	228	conflict_pgmap = get_dev_pagemap(PHYS_PFN(res->start), NULL);
15d36fec	229	if (conflict_pgmap) {
6869b7b2	230	WARN(1, "Conflicting mapping in same section\n");
15d36fec	231	put_dev_pagemap(conflict_pgmap);
50f44ee7 DW	232	error = -ENOMEM;
50f44ee7 DW	233	goto err_array;
15d36fec DJ	234	}
15d36fec DJ	235
7cc7867f	236	conflict_pgmap = get_dev_pagemap(PHYS_PFN(res->end), NULL);
15d36fec	237	if (conflict_pgmap) {
6869b7b2	238	WARN(1, "Conflicting mapping in same section\n");
15d36fec	239	put_dev_pagemap(conflict_pgmap);
50f44ee7 DW	240	error = -ENOMEM;
50f44ee7 DW	241	goto err_array;
15d36fec DJ	242	}
15d36fec DJ	243
7cc7867f	244	is_ram = region_intersects(res->start, resource_size(res),
d37a14bb	245	IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE);
41e94a85	246
06489cfb DW	247	if (is_ram != REGION_DISJOINT) {
	248	WARN_ONCE(1, "%s attempted on %s region %pr\n", __func__,
	249	is_ram == REGION_MIXED ? "mixed" : "ram", res);
a95c90f1 DW	250	error = -ENXIO;
a95c90f1 DW	251	goto err_array;
41e94a85 CH	252	}
41e94a85 CH	253
bcfa4b72 MW	254	error = xa_err(xa_store_range(&pgmap_array, PHYS_PFN(res->start),
bcfa4b72 MW	255	PHYS_PFN(res->end), pgmap, GFP_KERNEL));
9476df7d	256	if (error)
bcfa4b72	257	goto err_array;
9476df7d	258
41e94a85	259	if (nid < 0)
7eff93b7	260	nid = numa_mem_id();
41e94a85	261
7cc7867f DW	262	error = track_pfn_remap(NULL, &pgprot, PHYS_PFN(res->start), 0,
7cc7867f DW	263	resource_size(res));
9049771f DW	264	if (error)
	265	goto err_pfn_remap;
	266
f931ab47	267	mem_hotplug_begin();
69324b8f DW	268
	269	/*
	270	* For device private memory we call add_pages() as we only need to
	271	* allocate and initialize struct page for the device memory. More-
	272	* over the device memory is un-accessible thus we do not want to
	273	* create a linear mapping for the memory like arch_add_memory()
	274	* would do.
	275	*
	276	* For all other device memory types, which are accessible by
	277	* the CPU, we do want the linear mapping and thus use
	278	* arch_add_memory().
	279	*/
	280	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
7cc7867f DW	281	error = add_pages(nid, PHYS_PFN(res->start),
7cc7867f DW	282	PHYS_PFN(resource_size(res)), &restrictions);
69324b8f	283	} else {
7cc7867f	284	error = kasan_add_zero_shadow(__va(res->start), resource_size(res));
69324b8f DW	285	if (error) {
	286	mem_hotplug_done();
	287	goto err_kasan;
	288	}
	289
7cc7867f	290	error = arch_add_memory(nid, res->start, resource_size(res),
940519f0	291	&restrictions);
69324b8f DW	292	}
	293
	294	if (!error) {
	295	struct zone *zone;
	296
	297	zone = &NODE_DATA(nid)->node_zones[ZONE_DEVICE];
7cc7867f DW	298	move_pfn_range_to_zone(zone, PHYS_PFN(res->start),
7cc7867f DW	299	PHYS_PFN(resource_size(res)), restrictions.altmap);
0207df4f AR	300	}
0207df4f AR	301
f931ab47	302	mem_hotplug_done();
9476df7d DW	303	if (error)
9476df7d DW	304	goto err_add_memory;
41e94a85	305
966cf44f AD	306	/*
	307	* Initialization of the pages has been deferred until now in order
	308	* to allow us to do the work while not holding the hotplug lock.
	309	*/
	310	memmap_init_zone_device(&NODE_DATA(nid)->node_zones[ZONE_DEVICE],
7cc7867f DW	311	PHYS_PFN(res->start),
7cc7867f DW	312	PHYS_PFN(resource_size(res)), pgmap);
966cf44f	313	percpu_ref_get_many(pgmap->ref, pfn_end(pgmap) - pfn_first(pgmap));
41e94a85	314	return __va(res->start);
9476df7d DW	315
9476df7d DW	316	err_add_memory:
7cc7867f	317	kasan_remove_zero_shadow(__va(res->start), resource_size(res));
0207df4f	318	err_kasan:
7cc7867f	319	untrack_pfn(NULL, PHYS_PFN(res->start), resource_size(res));
9049771f	320	err_pfn_remap:
bcfa4b72 MW	321	pgmap_array_delete(res);
bcfa4b72 MW	322	err_array:
24917f6b CH	323	dev_pagemap_kill(pgmap);
24917f6b CH	324	dev_pagemap_cleanup(pgmap);
6f42193f	325	devmap_managed_enable_put();
9476df7d	326	return ERR_PTR(error);
41e94a85	327	}
6869b7b2 CH	328	EXPORT_SYMBOL_GPL(memremap_pages);
	329
	330	/**
	331	* devm_memremap_pages - remap and provide memmap backing for the given resource
	332	* @dev: hosting device for @res
	333	* @pgmap: pointer to a struct dev_pagemap
	334	*
	335	* Notes:
	336	* 1/ At a minimum the res and type members of @pgmap must be initialized
	337	* by the caller before passing it to this function
	338	*
	339	* 2/ The altmap field may optionally be initialized, in which case
	340	* PGMAP_ALTMAP_VALID must be set in pgmap->flags.
	341	*
	342	* 3/ The ref field may optionally be provided, in which pgmap->ref must be
	343	* 'live' on entry and will be killed and reaped at
	344	* devm_memremap_pages_release() time, or if this routine fails.
	345	*
	346	* 4/ res is expected to be a host memory range that could feasibly be
	347	* treated as a "System RAM" range, i.e. not a device mmio range, but
	348	* this is not enforced.
	349	*/
	350	void devm_memremap_pages(struct device dev, struct dev_pagemap *pgmap)
	351	{
	352	int error;
	353	void *ret;
	354
	355	ret = memremap_pages(pgmap, dev_to_node(dev));
	356	if (IS_ERR(ret))
	357	return ret;
	358
	359	error = devm_add_action_or_reset(dev, devm_memremap_pages_release,
	360	pgmap);
	361	if (error)
	362	return ERR_PTR(error);
	363	return ret;
	364	}
808153e1	365	EXPORT_SYMBOL_GPL(devm_memremap_pages);
4b94ffdc	366
2e3f139e DW	367	void devm_memunmap_pages(struct device dev, struct dev_pagemap pgmap)
	368	{
	369	devm_release_action(dev, devm_memremap_pages_release, pgmap);
	370	}
	371	EXPORT_SYMBOL_GPL(devm_memunmap_pages);
	372
4b94ffdc DW	373	unsigned long vmem_altmap_offset(struct vmem_altmap *altmap)
	374	{
	375	/* number of pfns from base where pfn_to_page() is valid */
514caf23 CH	376	if (altmap)
	377	return altmap->reserve + altmap->free;
	378	return 0;
4b94ffdc DW	379	}
	380
	381	void vmem_altmap_free(struct vmem_altmap *altmap, unsigned long nr_pfns)
	382	{
	383	altmap->alloc -= nr_pfns;
	384	}
	385
0822acb8 CH	386	/**
	387	* get_dev_pagemap() - take a new live reference on the dev_pagemap for @pfn
	388	* @pfn: page frame number to lookup page_map
	389	* @pgmap: optional known pgmap that already has a reference
	390	*
832d7aa0 CH	391	* If @pgmap is non-NULL and covers @pfn it will be returned as-is. If @pgmap
832d7aa0 CH	392	* is non-NULL but does not cover @pfn the reference to it will be released.
0822acb8 CH	393	*/
	394	struct dev_pagemap *get_dev_pagemap(unsigned long pfn,
	395	struct dev_pagemap *pgmap)
	396	{
0822acb8 CH	397	resource_size_t phys = PFN_PHYS(pfn);
	398
	399	/*
832d7aa0	400	* In the cached case we're already holding a live reference.
0822acb8	401	*/
832d7aa0	402	if (pgmap) {
e7744aa2	403	if (phys >= pgmap->res.start && phys <= pgmap->res.end)
832d7aa0 CH	404	return pgmap;
832d7aa0 CH	405	put_dev_pagemap(pgmap);
0822acb8 CH	406	}
	407
	408	/* fall back to slow path lookup */
	409	rcu_read_lock();
bcfa4b72	410	pgmap = xa_load(&pgmap_array, PHYS_PFN(phys));
0822acb8 CH	411	if (pgmap && !percpu_ref_tryget_live(pgmap->ref))
	412	pgmap = NULL;
	413	rcu_read_unlock();
	414
	415	return pgmap;
	416	}
e7638488	417	EXPORT_SYMBOL_GPL(get_dev_pagemap);
7b2d55d2	418
e7638488	419	#ifdef CONFIG_DEV_PAGEMAP_OPS
07d80269	420	void free_devmap_managed_page(struct page *page)
7b2d55d2	421	{
429589d6 DW	422	/* notify page idle for dax */
	423	if (!is_device_private_page(page)) {
	424	wake_up_var(&page->_refcount);
	425	return;
	426	}
7ab0ad0e	427
429589d6 DW	428	/* Clear Active bit in case of parallel mark_page_accessed */
	429	__ClearPageActive(page);
	430	__ClearPageWaiters(page);
	431
	432	mem_cgroup_uncharge(page);
	433
	434	/*
	435	* When a device_private page is freed, the page->mapping field
	436	* may still contain a (stale) mapping value. For example, the
	437	* lower bits of page->mapping may still identify the page as an
	438	* anonymous page. Ultimately, this entire field is just stale
	439	* and wrong, and it will cause errors if not cleared. One
	440	* example is:
	441	*
	442	* migrate_vma_pages()
	443	* migrate_vma_insert_page()
	444	* page_add_new_anon_rmap()
	445	* __page_set_anon_rmap()
	446	* ...checks page->mapping, via PageAnon(page) call,
	447	* and incorrectly concludes that the page is an
	448	* anonymous page. Therefore, it incorrectly,
	449	* silently fails to set up the new anon rmap.
	450	*
	451	* For other types of ZONE_DEVICE pages, migration is either
	452	* handled differently or not done at all, so there is no need
	453	* to clear page->mapping.
	454	*/
	455	page->mapping = NULL;
	456	page->pgmap->ops->page_free(page);
7b2d55d2	457	}
e7638488	458	#endif /* CONFIG_DEV_PAGEMAP_OPS */