[linux-2.6-block.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;
	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();
	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
		__remove_pages(page_zone(first_page), 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);
		}
		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;
77e080e7	106	struct page *first_page;
71389703	107	unsigned long pfn;
2c2a5af6	108	int nid;
71389703	109
24917f6b	110	dev_pagemap_kill(pgmap);
e7744aa2	111	for_each_device_pfn(pfn, pgmap)
71389703	112	put_page(pfn_to_page(pfn));
24917f6b	113	dev_pagemap_cleanup(pgmap);
9476df7d	114
77e080e7 AK	115	/* make sure to access a memmap that was actually initialized */
	116	first_page = pfn_to_page(pfn_first(pgmap));
	117
41e94a85	118	/* pages are dead and unused, undo the arch mapping */
77e080e7	119	nid = page_to_nid(first_page);
2c2a5af6	120
f931ab47	121	mem_hotplug_begin();
69324b8f	122	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
77e080e7 AK	123	__remove_pages(page_zone(first_page), PHYS_PFN(res->start),
77e080e7 AK	124	PHYS_PFN(resource_size(res)), NULL);
69324b8f	125	} else {
7cc7867f	126	arch_remove_memory(nid, res->start, resource_size(res),
514caf23	127	pgmap_altmap(pgmap));
7cc7867f	128	kasan_remove_zero_shadow(__va(res->start), resource_size(res));
69324b8f	129	}
f931ab47	130	mem_hotplug_done();
b5d24fda	131
7cc7867f	132	untrack_pfn(NULL, PHYS_PFN(res->start), resource_size(res));
bcfa4b72	133	pgmap_array_delete(res);
fdc029b1	134	WARN_ONCE(pgmap->altmap.alloc, "failed to free all reserved pages\n");
6f42193f	135	devmap_managed_enable_put();
9476df7d	136	}
6869b7b2 CH	137	EXPORT_SYMBOL_GPL(memunmap_pages);
	138
	139	static void devm_memremap_pages_release(void *data)
	140	{
	141	memunmap_pages(data);
	142	}
9476df7d	143
24917f6b CH	144	static void dev_pagemap_percpu_release(struct percpu_ref *ref)
	145	{
	146	struct dev_pagemap *pgmap =
	147	container_of(ref, struct dev_pagemap, internal_ref);
	148
	149	complete(&pgmap->done);
	150	}
	151
6869b7b2 CH	152	/*
	153	* Not device managed version of dev_memremap_pages, undone by
	154	* memunmap_pages(). Please use dev_memremap_pages if you have a struct
	155	* device available.
4b94ffdc	156	*/
6869b7b2	157	void memremap_pages(struct dev_pagemap pgmap, int nid)
41e94a85	158	{
949b9325	159	struct resource *res = &pgmap->res;
966cf44f	160	struct dev_pagemap *conflict_pgmap;
940519f0 MH	161	struct mhp_restrictions restrictions = {
	162	/*
	163	* We do not want any optional features only our own memmap
7cc7867f	164	*/
514caf23	165	.altmap = pgmap_altmap(pgmap),
940519f0	166	};
9049771f	167	pgprot_t pgprot = PAGE_KERNEL;
6869b7b2	168	int error, is_ram;
f6a55e1a	169	bool need_devmap_managed = true;
5f29a77c	170
3ed2dcdf CH	171	switch (pgmap->type) {
	172	case MEMORY_DEVICE_PRIVATE:
	173	if (!IS_ENABLED(CONFIG_DEVICE_PRIVATE)) {
	174	WARN(1, "Device private memory not supported\n");
	175	return ERR_PTR(-EINVAL);
	176	}
897e6365 CH	177	if (!pgmap->ops \|\| !pgmap->ops->migrate_to_ram) {
	178	WARN(1, "Missing migrate_to_ram method\n");
	179	return ERR_PTR(-EINVAL);
	180	}
3ed2dcdf CH	181	break;
	182	case MEMORY_DEVICE_FS_DAX:
	183	if (!IS_ENABLED(CONFIG_ZONE_DEVICE) \|\|
	184	IS_ENABLED(CONFIG_FS_DAX_LIMITED)) {
	185	WARN(1, "File system DAX not supported\n");
	186	return ERR_PTR(-EINVAL);
	187	}
	188	break;
	189	case MEMORY_DEVICE_DEVDAX:
	190	case MEMORY_DEVICE_PCI_P2PDMA:
f6a55e1a	191	need_devmap_managed = false;
3ed2dcdf CH	192	break;
	193	default:
	194	WARN(1, "Invalid pgmap type %d\n", pgmap->type);
	195	break;
	196	}
	197
24917f6b CH	198	if (!pgmap->ref) {
	199	if (pgmap->ops && (pgmap->ops->kill \|\| pgmap->ops->cleanup))
	200	return ERR_PTR(-EINVAL);
	201
	202	init_completion(&pgmap->done);
	203	error = percpu_ref_init(&pgmap->internal_ref,
	204	dev_pagemap_percpu_release, 0, GFP_KERNEL);
	205	if (error)
	206	return ERR_PTR(error);
	207	pgmap->ref = &pgmap->internal_ref;
	208	} else {
	209	if (!pgmap->ops \|\| !pgmap->ops->kill \|\| !pgmap->ops->cleanup) {
	210	WARN(1, "Missing reference count teardown definition\n");
	211	return ERR_PTR(-EINVAL);
	212	}
50f44ee7	213	}
a95c90f1	214
f6a55e1a	215	if (need_devmap_managed) {
6f42193f	216	error = devmap_managed_enable_get(pgmap);
f6a55e1a CH	217	if (error)
	218	return ERR_PTR(error);
	219	}
	220
7cc7867f	221	conflict_pgmap = get_dev_pagemap(PHYS_PFN(res->start), NULL);
15d36fec	222	if (conflict_pgmap) {
6869b7b2	223	WARN(1, "Conflicting mapping in same section\n");
15d36fec	224	put_dev_pagemap(conflict_pgmap);
50f44ee7 DW	225	error = -ENOMEM;
50f44ee7 DW	226	goto err_array;
15d36fec DJ	227	}
15d36fec DJ	228
7cc7867f	229	conflict_pgmap = get_dev_pagemap(PHYS_PFN(res->end), NULL);
15d36fec	230	if (conflict_pgmap) {
6869b7b2	231	WARN(1, "Conflicting mapping in same section\n");
15d36fec	232	put_dev_pagemap(conflict_pgmap);
50f44ee7 DW	233	error = -ENOMEM;
50f44ee7 DW	234	goto err_array;
15d36fec DJ	235	}
15d36fec DJ	236
7cc7867f	237	is_ram = region_intersects(res->start, resource_size(res),
d37a14bb	238	IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE);
41e94a85	239
06489cfb DW	240	if (is_ram != REGION_DISJOINT) {
	241	WARN_ONCE(1, "%s attempted on %s region %pr\n", __func__,
	242	is_ram == REGION_MIXED ? "mixed" : "ram", res);
a95c90f1 DW	243	error = -ENXIO;
a95c90f1 DW	244	goto err_array;
41e94a85 CH	245	}
41e94a85 CH	246
bcfa4b72 MW	247	error = xa_err(xa_store_range(&pgmap_array, PHYS_PFN(res->start),
bcfa4b72 MW	248	PHYS_PFN(res->end), pgmap, GFP_KERNEL));
9476df7d	249	if (error)
bcfa4b72	250	goto err_array;
9476df7d	251
41e94a85	252	if (nid < 0)
7eff93b7	253	nid = numa_mem_id();
41e94a85	254
7cc7867f DW	255	error = track_pfn_remap(NULL, &pgprot, PHYS_PFN(res->start), 0,
7cc7867f DW	256	resource_size(res));
9049771f DW	257	if (error)
	258	goto err_pfn_remap;
	259
f931ab47	260	mem_hotplug_begin();
69324b8f DW	261
	262	/*
	263	* For device private memory we call add_pages() as we only need to
	264	* allocate and initialize struct page for the device memory. More-
	265	* over the device memory is un-accessible thus we do not want to
	266	* create a linear mapping for the memory like arch_add_memory()
	267	* would do.
	268	*
	269	* For all other device memory types, which are accessible by
	270	* the CPU, we do want the linear mapping and thus use
	271	* arch_add_memory().
	272	*/
	273	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
7cc7867f DW	274	error = add_pages(nid, PHYS_PFN(res->start),
7cc7867f DW	275	PHYS_PFN(resource_size(res)), &restrictions);
69324b8f	276	} else {
7cc7867f	277	error = kasan_add_zero_shadow(__va(res->start), resource_size(res));
69324b8f DW	278	if (error) {
	279	mem_hotplug_done();
	280	goto err_kasan;
	281	}
	282
7cc7867f	283	error = arch_add_memory(nid, res->start, resource_size(res),
940519f0	284	&restrictions);
69324b8f DW	285	}
	286
	287	if (!error) {
	288	struct zone *zone;
	289
	290	zone = &NODE_DATA(nid)->node_zones[ZONE_DEVICE];
7cc7867f DW	291	move_pfn_range_to_zone(zone, PHYS_PFN(res->start),
7cc7867f DW	292	PHYS_PFN(resource_size(res)), restrictions.altmap);
0207df4f AR	293	}
0207df4f AR	294
f931ab47	295	mem_hotplug_done();
9476df7d DW	296	if (error)
9476df7d DW	297	goto err_add_memory;
41e94a85	298
966cf44f AD	299	/*
	300	* Initialization of the pages has been deferred until now in order
	301	* to allow us to do the work while not holding the hotplug lock.
	302	*/
	303	memmap_init_zone_device(&NODE_DATA(nid)->node_zones[ZONE_DEVICE],
7cc7867f DW	304	PHYS_PFN(res->start),
7cc7867f DW	305	PHYS_PFN(resource_size(res)), pgmap);
966cf44f	306	percpu_ref_get_many(pgmap->ref, pfn_end(pgmap) - pfn_first(pgmap));
41e94a85	307	return __va(res->start);
9476df7d DW	308
9476df7d DW	309	err_add_memory:
7cc7867f	310	kasan_remove_zero_shadow(__va(res->start), resource_size(res));
0207df4f	311	err_kasan:
7cc7867f	312	untrack_pfn(NULL, PHYS_PFN(res->start), resource_size(res));
9049771f	313	err_pfn_remap:
bcfa4b72 MW	314	pgmap_array_delete(res);
bcfa4b72 MW	315	err_array:
24917f6b CH	316	dev_pagemap_kill(pgmap);
24917f6b CH	317	dev_pagemap_cleanup(pgmap);
6f42193f	318	devmap_managed_enable_put();
9476df7d	319	return ERR_PTR(error);
41e94a85	320	}
6869b7b2 CH	321	EXPORT_SYMBOL_GPL(memremap_pages);
	322
	323	/**
	324	* devm_memremap_pages - remap and provide memmap backing for the given resource
	325	* @dev: hosting device for @res
	326	* @pgmap: pointer to a struct dev_pagemap
	327	*
	328	* Notes:
	329	* 1/ At a minimum the res and type members of @pgmap must be initialized
	330	* by the caller before passing it to this function
	331	*
	332	* 2/ The altmap field may optionally be initialized, in which case
	333	* PGMAP_ALTMAP_VALID must be set in pgmap->flags.
	334	*
	335	* 3/ The ref field may optionally be provided, in which pgmap->ref must be
	336	* 'live' on entry and will be killed and reaped at
	337	* devm_memremap_pages_release() time, or if this routine fails.
	338	*
	339	* 4/ res is expected to be a host memory range that could feasibly be
	340	* treated as a "System RAM" range, i.e. not a device mmio range, but
	341	* this is not enforced.
	342	*/
	343	void devm_memremap_pages(struct device dev, struct dev_pagemap *pgmap)
	344	{
	345	int error;
	346	void *ret;
	347
	348	ret = memremap_pages(pgmap, dev_to_node(dev));
	349	if (IS_ERR(ret))
	350	return ret;
	351
	352	error = devm_add_action_or_reset(dev, devm_memremap_pages_release,
	353	pgmap);
	354	if (error)
	355	return ERR_PTR(error);
	356	return ret;
	357	}
808153e1	358	EXPORT_SYMBOL_GPL(devm_memremap_pages);
4b94ffdc	359
2e3f139e DW	360	void devm_memunmap_pages(struct device dev, struct dev_pagemap pgmap)
	361	{
	362	devm_release_action(dev, devm_memremap_pages_release, pgmap);
	363	}
	364	EXPORT_SYMBOL_GPL(devm_memunmap_pages);
	365
4b94ffdc DW	366	unsigned long vmem_altmap_offset(struct vmem_altmap *altmap)
	367	{
	368	/* number of pfns from base where pfn_to_page() is valid */
514caf23 CH	369	if (altmap)
	370	return altmap->reserve + altmap->free;
	371	return 0;
4b94ffdc DW	372	}
	373
	374	void vmem_altmap_free(struct vmem_altmap *altmap, unsigned long nr_pfns)
	375	{
	376	altmap->alloc -= nr_pfns;
	377	}
	378
0822acb8 CH	379	/**
	380	* get_dev_pagemap() - take a new live reference on the dev_pagemap for @pfn
	381	* @pfn: page frame number to lookup page_map
	382	* @pgmap: optional known pgmap that already has a reference
	383	*
832d7aa0 CH	384	* If @pgmap is non-NULL and covers @pfn it will be returned as-is. If @pgmap
832d7aa0 CH	385	* is non-NULL but does not cover @pfn the reference to it will be released.
0822acb8 CH	386	*/
	387	struct dev_pagemap *get_dev_pagemap(unsigned long pfn,
	388	struct dev_pagemap *pgmap)
	389	{
0822acb8 CH	390	resource_size_t phys = PFN_PHYS(pfn);
	391
	392	/*
832d7aa0	393	* In the cached case we're already holding a live reference.
0822acb8	394	*/
832d7aa0	395	if (pgmap) {
e7744aa2	396	if (phys >= pgmap->res.start && phys <= pgmap->res.end)
832d7aa0 CH	397	return pgmap;
832d7aa0 CH	398	put_dev_pagemap(pgmap);
0822acb8 CH	399	}
	400
	401	/* fall back to slow path lookup */
	402	rcu_read_lock();
bcfa4b72	403	pgmap = xa_load(&pgmap_array, PHYS_PFN(phys));
0822acb8 CH	404	if (pgmap && !percpu_ref_tryget_live(pgmap->ref))
	405	pgmap = NULL;
	406	rcu_read_unlock();
	407
	408	return pgmap;
	409	}
e7638488	410	EXPORT_SYMBOL_GPL(get_dev_pagemap);
7b2d55d2	411
e7638488	412	#ifdef CONFIG_DEV_PAGEMAP_OPS
e7638488	413	void __put_devmap_managed_page(struct page *page)
7b2d55d2 JG	414	{
	415	int count = page_ref_dec_return(page);
	416
	417	/*
	418	* If refcount is 1 then page is freed and refcount is stable as nobody
	419	* holds a reference on the page.
	420	*/
	421	if (count == 1) {
	422	/* Clear Active bit in case of parallel mark_page_accessed */
	423	__ClearPageActive(page);
	424	__ClearPageWaiters(page);
	425
c733a828	426	mem_cgroup_uncharge(page);
7b2d55d2	427
7ab0ad0e RC	428	/*
	429	* When a device_private page is freed, the page->mapping field
	430	* may still contain a (stale) mapping value. For example, the
	431	* lower bits of page->mapping may still identify the page as
	432	* an anonymous page. Ultimately, this entire field is just
	433	* stale and wrong, and it will cause errors if not cleared.
	434	* One example is:
	435	*
	436	* migrate_vma_pages()
	437	* migrate_vma_insert_page()
	438	* page_add_new_anon_rmap()
	439	* __page_set_anon_rmap()
	440	* ...checks page->mapping, via PageAnon(page) call,
	441	* and incorrectly concludes that the page is an
	442	* anonymous page. Therefore, it incorrectly,
	443	* silently fails to set up the new anon rmap.
	444	*
	445	* For other types of ZONE_DEVICE pages, migration is either
	446	* handled differently or not done at all, so there is no need
	447	* to clear page->mapping.
	448	*/
	449	if (is_device_private_page(page))
	450	page->mapping = NULL;
	451
80a72d0a	452	page->pgmap->ops->page_free(page);
7b2d55d2 JG	453	} else if (!count)
	454	__put_page(page);
	455	}
31c5bda3	456	EXPORT_SYMBOL(__put_devmap_managed_page);
e7638488	457	#endif /* CONFIG_DEV_PAGEMAP_OPS */