[mirror_ubuntu-hirsute-kernel.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->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;
	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);

	/* pages are dead and unused, undo the arch mapping */
	nid = page_to_nid(pfn_to_page(PHYS_PFN(res->start)));

	mem_hotplug_begin();
	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
		pfn = PHYS_PFN(res->start);
		__remove_pages(page_zone(pfn_to_page(pfn)), pfn,
				 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);
		}
		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 __put_devmap_managed_page(struct page *page)
{
	int count = page_ref_dec_return(page);

	/*
	 * If refcount is 1 then page is freed and refcount is stable as nobody
	 * holds a reference on the page.
	 */
	if (count == 1) {
		/* 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.
		 */
		if (is_device_private_page(page))
			page->mapping = NULL;

		page->pgmap->ops->page_free(page);
	} else if (!count)
		__put_page(page);
}
EXPORT_SYMBOL(__put_devmap_managed_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	{
24917f6b	30	if (!pgmap->ops \|\| !pgmap->ops->page_free) {
f6a55e1a CH	31	WARN(1, "Missing page_free method\n");
	32	return -EINVAL;
	33	}
	34
	35	if (atomic_inc_return(&devmap_managed_enable) == 1)
	36	static_branch_enable(&devmap_managed_key);
6f42193f	37	return 0;
f6a55e1a CH	38	}
f6a55e1a CH	39	#else
6f42193f	40	static int devmap_managed_enable_get(struct dev_pagemap *pgmap)
f6a55e1a CH	41	{
	42	return -EINVAL;
	43	}
6f42193f CH	44	static void devmap_managed_enable_put(void)
	45	{
	46	}
f6a55e1a CH	47	#endif /* CONFIG_DEV_PAGEMAP_OPS */
f6a55e1a CH	48
bcfa4b72	49	static void pgmap_array_delete(struct resource *res)
ab1b597e	50	{
bcfa4b72 MW	51	xa_store_range(&pgmap_array, PHYS_PFN(res->start), PHYS_PFN(res->end),
bcfa4b72 MW	52	NULL, GFP_KERNEL);
ab1b597e	53	synchronize_rcu();
9476df7d DW	54	}
9476df7d DW	55
e7744aa2	56	static unsigned long pfn_first(struct dev_pagemap *pgmap)
5c2c2587	57	{
7cc7867f	58	return PHYS_PFN(pgmap->res.start) +
514caf23	59	vmem_altmap_offset(pgmap_altmap(pgmap));
5c2c2587 DW	60	}
5c2c2587 DW	61
e7744aa2	62	static unsigned long pfn_end(struct dev_pagemap *pgmap)
5c2c2587	63	{
e7744aa2	64	const struct resource *res = &pgmap->res;
5c2c2587 DW	65
	66	return (res->start + resource_size(res)) >> PAGE_SHIFT;
	67	}
	68
949b9325 DW	69	static unsigned long pfn_next(unsigned long pfn)
	70	{
	71	if (pfn % 1024 == 0)
	72	cond_resched();
	73	return pfn + 1;
	74	}
	75
5c2c2587	76	#define for_each_device_pfn(pfn, map) \
949b9325	77	for (pfn = pfn_first(map); pfn < pfn_end(map); pfn = pfn_next(pfn))
5c2c2587	78
24917f6b CH	79	static void dev_pagemap_kill(struct dev_pagemap *pgmap)
	80	{
	81	if (pgmap->ops && pgmap->ops->kill)
	82	pgmap->ops->kill(pgmap);
	83	else
	84	percpu_ref_kill(pgmap->ref);
	85	}
	86
	87	static void dev_pagemap_cleanup(struct dev_pagemap *pgmap)
	88	{
	89	if (pgmap->ops && pgmap->ops->cleanup) {
	90	pgmap->ops->cleanup(pgmap);
	91	} else {
	92	wait_for_completion(&pgmap->done);
	93	percpu_ref_exit(pgmap->ref);
	94	}
06282373 DW	95	/*
	96	* Undo the pgmap ref assignment for the internal case as the
	97	* caller may re-enable the same pgmap.
	98	*/
	99	if (pgmap->ref == &pgmap->internal_ref)
	100	pgmap->ref = NULL;
24917f6b CH	101	}
24917f6b CH	102
6869b7b2	103	void memunmap_pages(struct dev_pagemap *pgmap)
41e94a85	104	{
e7744aa2	105	struct resource *res = &pgmap->res;
71389703	106	unsigned long pfn;
2c2a5af6	107	int nid;
71389703	108
24917f6b	109	dev_pagemap_kill(pgmap);
e7744aa2	110	for_each_device_pfn(pfn, pgmap)
71389703	111	put_page(pfn_to_page(pfn));
24917f6b	112	dev_pagemap_cleanup(pgmap);
9476df7d	113
41e94a85	114	/* pages are dead and unused, undo the arch mapping */
7cc7867f	115	nid = page_to_nid(pfn_to_page(PHYS_PFN(res->start)));
2c2a5af6	116
f931ab47	117	mem_hotplug_begin();
69324b8f	118	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
7cc7867f	119	pfn = PHYS_PFN(res->start);
69324b8f	120	__remove_pages(page_zone(pfn_to_page(pfn)), pfn,
7cc7867f	121	PHYS_PFN(resource_size(res)), NULL);
69324b8f	122	} else {
7cc7867f	123	arch_remove_memory(nid, res->start, resource_size(res),
514caf23	124	pgmap_altmap(pgmap));
7cc7867f	125	kasan_remove_zero_shadow(__va(res->start), resource_size(res));
69324b8f	126	}
f931ab47	127	mem_hotplug_done();
b5d24fda	128
7cc7867f	129	untrack_pfn(NULL, PHYS_PFN(res->start), resource_size(res));
bcfa4b72	130	pgmap_array_delete(res);
fdc029b1	131	WARN_ONCE(pgmap->altmap.alloc, "failed to free all reserved pages\n");
6f42193f	132	devmap_managed_enable_put();
9476df7d	133	}
6869b7b2 CH	134	EXPORT_SYMBOL_GPL(memunmap_pages);
	135
	136	static void devm_memremap_pages_release(void *data)
	137	{
	138	memunmap_pages(data);
	139	}
9476df7d	140
24917f6b CH	141	static void dev_pagemap_percpu_release(struct percpu_ref *ref)
	142	{
	143	struct dev_pagemap *pgmap =
	144	container_of(ref, struct dev_pagemap, internal_ref);
	145
	146	complete(&pgmap->done);
	147	}
	148
6869b7b2 CH	149	/*
	150	* Not device managed version of dev_memremap_pages, undone by
	151	* memunmap_pages(). Please use dev_memremap_pages if you have a struct
	152	* device available.
4b94ffdc	153	*/
6869b7b2	154	void memremap_pages(struct dev_pagemap pgmap, int nid)
41e94a85	155	{
949b9325	156	struct resource *res = &pgmap->res;
966cf44f	157	struct dev_pagemap *conflict_pgmap;
940519f0 MH	158	struct mhp_restrictions restrictions = {
	159	/*
	160	* We do not want any optional features only our own memmap
7cc7867f	161	*/
514caf23	162	.altmap = pgmap_altmap(pgmap),
940519f0	163	};
9049771f	164	pgprot_t pgprot = PAGE_KERNEL;
6869b7b2	165	int error, is_ram;
f6a55e1a	166	bool need_devmap_managed = true;
5f29a77c	167
3ed2dcdf CH	168	switch (pgmap->type) {
	169	case MEMORY_DEVICE_PRIVATE:
	170	if (!IS_ENABLED(CONFIG_DEVICE_PRIVATE)) {
	171	WARN(1, "Device private memory not supported\n");
	172	return ERR_PTR(-EINVAL);
	173	}
897e6365 CH	174	if (!pgmap->ops \|\| !pgmap->ops->migrate_to_ram) {
	175	WARN(1, "Missing migrate_to_ram method\n");
	176	return ERR_PTR(-EINVAL);
	177	}
3ed2dcdf CH	178	break;
	179	case MEMORY_DEVICE_FS_DAX:
	180	if (!IS_ENABLED(CONFIG_ZONE_DEVICE) \|\|
	181	IS_ENABLED(CONFIG_FS_DAX_LIMITED)) {
	182	WARN(1, "File system DAX not supported\n");
	183	return ERR_PTR(-EINVAL);
	184	}
	185	break;
	186	case MEMORY_DEVICE_DEVDAX:
	187	case MEMORY_DEVICE_PCI_P2PDMA:
f6a55e1a	188	need_devmap_managed = false;
3ed2dcdf CH	189	break;
	190	default:
	191	WARN(1, "Invalid pgmap type %d\n", pgmap->type);
	192	break;
	193	}
	194
24917f6b CH	195	if (!pgmap->ref) {
	196	if (pgmap->ops && (pgmap->ops->kill \|\| pgmap->ops->cleanup))
	197	return ERR_PTR(-EINVAL);
	198
	199	init_completion(&pgmap->done);
	200	error = percpu_ref_init(&pgmap->internal_ref,
	201	dev_pagemap_percpu_release, 0, GFP_KERNEL);
	202	if (error)
	203	return ERR_PTR(error);
	204	pgmap->ref = &pgmap->internal_ref;
	205	} else {
	206	if (!pgmap->ops \|\| !pgmap->ops->kill \|\| !pgmap->ops->cleanup) {
	207	WARN(1, "Missing reference count teardown definition\n");
	208	return ERR_PTR(-EINVAL);
	209	}
50f44ee7	210	}
a95c90f1	211
f6a55e1a	212	if (need_devmap_managed) {
6f42193f	213	error = devmap_managed_enable_get(pgmap);
f6a55e1a CH	214	if (error)
	215	return ERR_PTR(error);
	216	}
	217
7cc7867f	218	conflict_pgmap = get_dev_pagemap(PHYS_PFN(res->start), NULL);
15d36fec	219	if (conflict_pgmap) {
6869b7b2	220	WARN(1, "Conflicting mapping in same section\n");
15d36fec	221	put_dev_pagemap(conflict_pgmap);
50f44ee7 DW	222	error = -ENOMEM;
50f44ee7 DW	223	goto err_array;
15d36fec DJ	224	}
15d36fec DJ	225
7cc7867f	226	conflict_pgmap = get_dev_pagemap(PHYS_PFN(res->end), NULL);
15d36fec	227	if (conflict_pgmap) {
6869b7b2	228	WARN(1, "Conflicting mapping in same section\n");
15d36fec	229	put_dev_pagemap(conflict_pgmap);
50f44ee7 DW	230	error = -ENOMEM;
50f44ee7 DW	231	goto err_array;
15d36fec DJ	232	}
15d36fec DJ	233
7cc7867f	234	is_ram = region_intersects(res->start, resource_size(res),
d37a14bb	235	IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE);
41e94a85	236
06489cfb DW	237	if (is_ram != REGION_DISJOINT) {
	238	WARN_ONCE(1, "%s attempted on %s region %pr\n", __func__,
	239	is_ram == REGION_MIXED ? "mixed" : "ram", res);
a95c90f1 DW	240	error = -ENXIO;
a95c90f1 DW	241	goto err_array;
41e94a85 CH	242	}
41e94a85 CH	243
bcfa4b72 MW	244	error = xa_err(xa_store_range(&pgmap_array, PHYS_PFN(res->start),
bcfa4b72 MW	245	PHYS_PFN(res->end), pgmap, GFP_KERNEL));
9476df7d	246	if (error)
bcfa4b72	247	goto err_array;
9476df7d	248
41e94a85	249	if (nid < 0)
7eff93b7	250	nid = numa_mem_id();
41e94a85	251
7cc7867f DW	252	error = track_pfn_remap(NULL, &pgprot, PHYS_PFN(res->start), 0,
7cc7867f DW	253	resource_size(res));
9049771f DW	254	if (error)
	255	goto err_pfn_remap;
	256
f931ab47	257	mem_hotplug_begin();
69324b8f DW	258
	259	/*
	260	* For device private memory we call add_pages() as we only need to
	261	* allocate and initialize struct page for the device memory. More-
	262	* over the device memory is un-accessible thus we do not want to
	263	* create a linear mapping for the memory like arch_add_memory()
	264	* would do.
	265	*
	266	* For all other device memory types, which are accessible by
	267	* the CPU, we do want the linear mapping and thus use
	268	* arch_add_memory().
	269	*/
	270	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
7cc7867f DW	271	error = add_pages(nid, PHYS_PFN(res->start),
7cc7867f DW	272	PHYS_PFN(resource_size(res)), &restrictions);
69324b8f	273	} else {
7cc7867f	274	error = kasan_add_zero_shadow(__va(res->start), resource_size(res));
69324b8f DW	275	if (error) {
	276	mem_hotplug_done();
	277	goto err_kasan;
	278	}
	279
7cc7867f	280	error = arch_add_memory(nid, res->start, resource_size(res),
940519f0	281	&restrictions);
69324b8f DW	282	}
	283
	284	if (!error) {
	285	struct zone *zone;
	286
	287	zone = &NODE_DATA(nid)->node_zones[ZONE_DEVICE];
7cc7867f DW	288	move_pfn_range_to_zone(zone, PHYS_PFN(res->start),
7cc7867f DW	289	PHYS_PFN(resource_size(res)), restrictions.altmap);
0207df4f AR	290	}
0207df4f AR	291
f931ab47	292	mem_hotplug_done();
9476df7d DW	293	if (error)
9476df7d DW	294	goto err_add_memory;
41e94a85	295
966cf44f AD	296	/*
	297	* Initialization of the pages has been deferred until now in order
	298	* to allow us to do the work while not holding the hotplug lock.
	299	*/
	300	memmap_init_zone_device(&NODE_DATA(nid)->node_zones[ZONE_DEVICE],
7cc7867f DW	301	PHYS_PFN(res->start),
7cc7867f DW	302	PHYS_PFN(resource_size(res)), pgmap);
966cf44f	303	percpu_ref_get_many(pgmap->ref, pfn_end(pgmap) - pfn_first(pgmap));
41e94a85	304	return __va(res->start);
9476df7d DW	305
9476df7d DW	306	err_add_memory:
7cc7867f	307	kasan_remove_zero_shadow(__va(res->start), resource_size(res));
0207df4f	308	err_kasan:
7cc7867f	309	untrack_pfn(NULL, PHYS_PFN(res->start), resource_size(res));
9049771f	310	err_pfn_remap:
bcfa4b72 MW	311	pgmap_array_delete(res);
bcfa4b72 MW	312	err_array:
24917f6b CH	313	dev_pagemap_kill(pgmap);
24917f6b CH	314	dev_pagemap_cleanup(pgmap);
6f42193f	315	devmap_managed_enable_put();
9476df7d	316	return ERR_PTR(error);
41e94a85	317	}
6869b7b2 CH	318	EXPORT_SYMBOL_GPL(memremap_pages);
	319
	320	/**
	321	* devm_memremap_pages - remap and provide memmap backing for the given resource
	322	* @dev: hosting device for @res
	323	* @pgmap: pointer to a struct dev_pagemap
	324	*
	325	* Notes:
	326	* 1/ At a minimum the res and type members of @pgmap must be initialized
	327	* by the caller before passing it to this function
	328	*
	329	* 2/ The altmap field may optionally be initialized, in which case
	330	* PGMAP_ALTMAP_VALID must be set in pgmap->flags.
	331	*
	332	* 3/ The ref field may optionally be provided, in which pgmap->ref must be
	333	* 'live' on entry and will be killed and reaped at
	334	* devm_memremap_pages_release() time, or if this routine fails.
	335	*
	336	* 4/ res is expected to be a host memory range that could feasibly be
	337	* treated as a "System RAM" range, i.e. not a device mmio range, but
	338	* this is not enforced.
	339	*/
	340	void devm_memremap_pages(struct device dev, struct dev_pagemap *pgmap)
	341	{
	342	int error;
	343	void *ret;
	344
	345	ret = memremap_pages(pgmap, dev_to_node(dev));
	346	if (IS_ERR(ret))
	347	return ret;
	348
	349	error = devm_add_action_or_reset(dev, devm_memremap_pages_release,
	350	pgmap);
	351	if (error)
	352	return ERR_PTR(error);
	353	return ret;
	354	}
808153e1	355	EXPORT_SYMBOL_GPL(devm_memremap_pages);
4b94ffdc	356
2e3f139e DW	357	void devm_memunmap_pages(struct device dev, struct dev_pagemap pgmap)
	358	{
	359	devm_release_action(dev, devm_memremap_pages_release, pgmap);
	360	}
	361	EXPORT_SYMBOL_GPL(devm_memunmap_pages);
	362
4b94ffdc DW	363	unsigned long vmem_altmap_offset(struct vmem_altmap *altmap)
	364	{
	365	/* number of pfns from base where pfn_to_page() is valid */
514caf23 CH	366	if (altmap)
	367	return altmap->reserve + altmap->free;
	368	return 0;
4b94ffdc DW	369	}
	370
	371	void vmem_altmap_free(struct vmem_altmap *altmap, unsigned long nr_pfns)
	372	{
	373	altmap->alloc -= nr_pfns;
	374	}
	375
0822acb8 CH	376	/**
	377	* get_dev_pagemap() - take a new live reference on the dev_pagemap for @pfn
	378	* @pfn: page frame number to lookup page_map
	379	* @pgmap: optional known pgmap that already has a reference
	380	*
832d7aa0 CH	381	* If @pgmap is non-NULL and covers @pfn it will be returned as-is. If @pgmap
832d7aa0 CH	382	* is non-NULL but does not cover @pfn the reference to it will be released.
0822acb8 CH	383	*/
	384	struct dev_pagemap *get_dev_pagemap(unsigned long pfn,
	385	struct dev_pagemap *pgmap)
	386	{
0822acb8 CH	387	resource_size_t phys = PFN_PHYS(pfn);
	388
	389	/*
832d7aa0	390	* In the cached case we're already holding a live reference.
0822acb8	391	*/
832d7aa0	392	if (pgmap) {
e7744aa2	393	if (phys >= pgmap->res.start && phys <= pgmap->res.end)
832d7aa0 CH	394	return pgmap;
832d7aa0 CH	395	put_dev_pagemap(pgmap);
0822acb8 CH	396	}
	397
	398	/* fall back to slow path lookup */
	399	rcu_read_lock();
bcfa4b72	400	pgmap = xa_load(&pgmap_array, PHYS_PFN(phys));
0822acb8 CH	401	if (pgmap && !percpu_ref_tryget_live(pgmap->ref))
	402	pgmap = NULL;
	403	rcu_read_unlock();
	404
	405	return pgmap;
	406	}
e7638488	407	EXPORT_SYMBOL_GPL(get_dev_pagemap);
7b2d55d2	408
e7638488	409	#ifdef CONFIG_DEV_PAGEMAP_OPS
e7638488	410	void __put_devmap_managed_page(struct page *page)
7b2d55d2 JG	411	{
	412	int count = page_ref_dec_return(page);
	413
	414	/*
	415	* If refcount is 1 then page is freed and refcount is stable as nobody
	416	* holds a reference on the page.
	417	*/
	418	if (count == 1) {
	419	/* Clear Active bit in case of parallel mark_page_accessed */
	420	__ClearPageActive(page);
	421	__ClearPageWaiters(page);
	422
c733a828	423	mem_cgroup_uncharge(page);
7b2d55d2	424
7ab0ad0e RC	425	/*
	426	* When a device_private page is freed, the page->mapping field
	427	* may still contain a (stale) mapping value. For example, the
	428	* lower bits of page->mapping may still identify the page as
	429	* an anonymous page. Ultimately, this entire field is just
	430	* stale and wrong, and it will cause errors if not cleared.
	431	* One example is:
	432	*
	433	* migrate_vma_pages()
	434	* migrate_vma_insert_page()
	435	* page_add_new_anon_rmap()
	436	* __page_set_anon_rmap()
	437	* ...checks page->mapping, via PageAnon(page) call,
	438	* and incorrectly concludes that the page is an
	439	* anonymous page. Therefore, it incorrectly,
	440	* silently fails to set up the new anon rmap.
	441	*
	442	* For other types of ZONE_DEVICE pages, migration is either
	443	* handled differently or not done at all, so there is no need
	444	* to clear page->mapping.
	445	*/
	446	if (is_device_private_page(page))
	447	page->mapping = NULL;
	448
80a72d0a	449	page->pgmap->ops->page_free(page);
7b2d55d2 JG	450	} else if (!count)
	451	__put_page(page);
	452	}
31c5bda3	453	EXPORT_SYMBOL(__put_devmap_managed_page);
e7638488	454	#endif /* CONFIG_DEV_PAGEMAP_OPS */