[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);
#define SECTION_MASK ~((1UL << PA_SECTION_SHIFT) - 1)
#define SECTION_SIZE (1UL << PA_SECTION_SHIFT)

#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 DW	16	#define SECTION_MASK ~((1UL << PA_SECTION_SHIFT) - 1)
	17	#define SECTION_SIZE (1UL << PA_SECTION_SHIFT)
	18
f6a55e1a CH	19	#ifdef CONFIG_DEV_PAGEMAP_OPS
	20	DEFINE_STATIC_KEY_FALSE(devmap_managed_key);
	21	EXPORT_SYMBOL(devmap_managed_key);
	22	static atomic_t devmap_managed_enable;
	23
6f42193f	24	static void devmap_managed_enable_put(void)
f6a55e1a CH	25	{
	26	if (atomic_dec_and_test(&devmap_managed_enable))
	27	static_branch_disable(&devmap_managed_key);
	28	}
	29
6f42193f	30	static int devmap_managed_enable_get(struct dev_pagemap *pgmap)
f6a55e1a	31	{
24917f6b	32	if (!pgmap->ops \|\| !pgmap->ops->page_free) {
f6a55e1a CH	33	WARN(1, "Missing page_free method\n");
	34	return -EINVAL;
	35	}
	36
	37	if (atomic_inc_return(&devmap_managed_enable) == 1)
	38	static_branch_enable(&devmap_managed_key);
6f42193f	39	return 0;
f6a55e1a CH	40	}
f6a55e1a CH	41	#else
6f42193f	42	static int devmap_managed_enable_get(struct dev_pagemap *pgmap)
f6a55e1a CH	43	{
	44	return -EINVAL;
	45	}
6f42193f CH	46	static void devmap_managed_enable_put(void)
	47	{
	48	}
f6a55e1a CH	49	#endif /* CONFIG_DEV_PAGEMAP_OPS */
f6a55e1a CH	50
bcfa4b72	51	static void pgmap_array_delete(struct resource *res)
ab1b597e	52	{
bcfa4b72 MW	53	xa_store_range(&pgmap_array, PHYS_PFN(res->start), PHYS_PFN(res->end),
bcfa4b72 MW	54	NULL, GFP_KERNEL);
ab1b597e	55	synchronize_rcu();
9476df7d DW	56	}
9476df7d DW	57
e7744aa2	58	static unsigned long pfn_first(struct dev_pagemap *pgmap)
5c2c2587	59	{
7cc7867f	60	return PHYS_PFN(pgmap->res.start) +
514caf23	61	vmem_altmap_offset(pgmap_altmap(pgmap));
5c2c2587 DW	62	}
5c2c2587 DW	63
e7744aa2	64	static unsigned long pfn_end(struct dev_pagemap *pgmap)
5c2c2587	65	{
e7744aa2	66	const struct resource *res = &pgmap->res;
5c2c2587 DW	67
	68	return (res->start + resource_size(res)) >> PAGE_SHIFT;
	69	}
	70
949b9325 DW	71	static unsigned long pfn_next(unsigned long pfn)
	72	{
	73	if (pfn % 1024 == 0)
	74	cond_resched();
	75	return pfn + 1;
	76	}
	77
5c2c2587	78	#define for_each_device_pfn(pfn, map) \
949b9325	79	for (pfn = pfn_first(map); pfn < pfn_end(map); pfn = pfn_next(pfn))
5c2c2587	80
24917f6b CH	81	static void dev_pagemap_kill(struct dev_pagemap *pgmap)
	82	{
	83	if (pgmap->ops && pgmap->ops->kill)
	84	pgmap->ops->kill(pgmap);
	85	else
	86	percpu_ref_kill(pgmap->ref);
	87	}
	88
	89	static void dev_pagemap_cleanup(struct dev_pagemap *pgmap)
	90	{
	91	if (pgmap->ops && pgmap->ops->cleanup) {
	92	pgmap->ops->cleanup(pgmap);
	93	} else {
	94	wait_for_completion(&pgmap->done);
	95	percpu_ref_exit(pgmap->ref);
	96	}
06282373 DW	97	/*
	98	* Undo the pgmap ref assignment for the internal case as the
	99	* caller may re-enable the same pgmap.
	100	*/
	101	if (pgmap->ref == &pgmap->internal_ref)
	102	pgmap->ref = NULL;
24917f6b CH	103	}
24917f6b CH	104
6869b7b2	105	void memunmap_pages(struct dev_pagemap *pgmap)
41e94a85	106	{
e7744aa2	107	struct resource *res = &pgmap->res;
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
41e94a85	116	/* pages are dead and unused, undo the arch mapping */
7cc7867f	117	nid = page_to_nid(pfn_to_page(PHYS_PFN(res->start)));
2c2a5af6	118
f931ab47	119	mem_hotplug_begin();
69324b8f	120	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
7cc7867f	121	pfn = PHYS_PFN(res->start);
69324b8f	122	__remove_pages(page_zone(pfn_to_page(pfn)), pfn,
7cc7867f	123	PHYS_PFN(resource_size(res)), NULL);
69324b8f	124	} else {
7cc7867f	125	arch_remove_memory(nid, res->start, resource_size(res),
514caf23	126	pgmap_altmap(pgmap));
7cc7867f	127	kasan_remove_zero_shadow(__va(res->start), resource_size(res));
69324b8f	128	}
f931ab47	129	mem_hotplug_done();
b5d24fda	130
7cc7867f	131	untrack_pfn(NULL, PHYS_PFN(res->start), resource_size(res));
bcfa4b72	132	pgmap_array_delete(res);
fdc029b1	133	WARN_ONCE(pgmap->altmap.alloc, "failed to free all reserved pages\n");
6f42193f	134	devmap_managed_enable_put();
9476df7d	135	}
6869b7b2 CH	136	EXPORT_SYMBOL_GPL(memunmap_pages);
	137
	138	static void devm_memremap_pages_release(void *data)
	139	{
	140	memunmap_pages(data);
	141	}
9476df7d	142
24917f6b CH	143	static void dev_pagemap_percpu_release(struct percpu_ref *ref)
	144	{
	145	struct dev_pagemap *pgmap =
	146	container_of(ref, struct dev_pagemap, internal_ref);
	147
	148	complete(&pgmap->done);
	149	}
	150
6869b7b2 CH	151	/*
	152	* Not device managed version of dev_memremap_pages, undone by
	153	* memunmap_pages(). Please use dev_memremap_pages if you have a struct
	154	* device available.
4b94ffdc	155	*/
6869b7b2	156	void memremap_pages(struct dev_pagemap pgmap, int nid)
41e94a85	157	{
949b9325	158	struct resource *res = &pgmap->res;
966cf44f	159	struct dev_pagemap *conflict_pgmap;
940519f0 MH	160	struct mhp_restrictions restrictions = {
	161	/*
	162	* We do not want any optional features only our own memmap
7cc7867f	163	*/
514caf23	164	.altmap = pgmap_altmap(pgmap),
940519f0	165	};
9049771f	166	pgprot_t pgprot = PAGE_KERNEL;
6869b7b2	167	int error, is_ram;
f6a55e1a	168	bool need_devmap_managed = true;
5f29a77c	169
3ed2dcdf CH	170	switch (pgmap->type) {
	171	case MEMORY_DEVICE_PRIVATE:
	172	if (!IS_ENABLED(CONFIG_DEVICE_PRIVATE)) {
	173	WARN(1, "Device private memory not supported\n");
	174	return ERR_PTR(-EINVAL);
	175	}
897e6365 CH	176	if (!pgmap->ops \|\| !pgmap->ops->migrate_to_ram) {
	177	WARN(1, "Missing migrate_to_ram method\n");
	178	return ERR_PTR(-EINVAL);
	179	}
3ed2dcdf CH	180	break;
	181	case MEMORY_DEVICE_FS_DAX:
	182	if (!IS_ENABLED(CONFIG_ZONE_DEVICE) \|\|
	183	IS_ENABLED(CONFIG_FS_DAX_LIMITED)) {
	184	WARN(1, "File system DAX not supported\n");
	185	return ERR_PTR(-EINVAL);
	186	}
	187	break;
	188	case MEMORY_DEVICE_DEVDAX:
	189	case MEMORY_DEVICE_PCI_P2PDMA:
f6a55e1a	190	need_devmap_managed = false;
3ed2dcdf CH	191	break;
	192	default:
	193	WARN(1, "Invalid pgmap type %d\n", pgmap->type);
	194	break;
	195	}
	196
24917f6b CH	197	if (!pgmap->ref) {
	198	if (pgmap->ops && (pgmap->ops->kill \|\| pgmap->ops->cleanup))
	199	return ERR_PTR(-EINVAL);
	200
	201	init_completion(&pgmap->done);
	202	error = percpu_ref_init(&pgmap->internal_ref,
	203	dev_pagemap_percpu_release, 0, GFP_KERNEL);
	204	if (error)
	205	return ERR_PTR(error);
	206	pgmap->ref = &pgmap->internal_ref;
	207	} else {
	208	if (!pgmap->ops \|\| !pgmap->ops->kill \|\| !pgmap->ops->cleanup) {
	209	WARN(1, "Missing reference count teardown definition\n");
	210	return ERR_PTR(-EINVAL);
	211	}
50f44ee7	212	}
a95c90f1	213
f6a55e1a	214	if (need_devmap_managed) {
6f42193f	215	error = devmap_managed_enable_get(pgmap);
f6a55e1a CH	216	if (error)
	217	return ERR_PTR(error);
	218	}
	219
7cc7867f	220	conflict_pgmap = get_dev_pagemap(PHYS_PFN(res->start), NULL);
15d36fec	221	if (conflict_pgmap) {
6869b7b2	222	WARN(1, "Conflicting mapping in same section\n");
15d36fec	223	put_dev_pagemap(conflict_pgmap);
50f44ee7 DW	224	error = -ENOMEM;
50f44ee7 DW	225	goto err_array;
15d36fec DJ	226	}
15d36fec DJ	227
7cc7867f	228	conflict_pgmap = get_dev_pagemap(PHYS_PFN(res->end), 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	is_ram = region_intersects(res->start, resource_size(res),
d37a14bb	237	IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE);
41e94a85	238
06489cfb DW	239	if (is_ram != REGION_DISJOINT) {
	240	WARN_ONCE(1, "%s attempted on %s region %pr\n", __func__,
	241	is_ram == REGION_MIXED ? "mixed" : "ram", res);
a95c90f1 DW	242	error = -ENXIO;
a95c90f1 DW	243	goto err_array;
41e94a85 CH	244	}
41e94a85 CH	245
bcfa4b72 MW	246	error = xa_err(xa_store_range(&pgmap_array, PHYS_PFN(res->start),
bcfa4b72 MW	247	PHYS_PFN(res->end), pgmap, GFP_KERNEL));
9476df7d	248	if (error)
bcfa4b72	249	goto err_array;
9476df7d	250
41e94a85	251	if (nid < 0)
7eff93b7	252	nid = numa_mem_id();
41e94a85	253
7cc7867f DW	254	error = track_pfn_remap(NULL, &pgprot, PHYS_PFN(res->start), 0,
7cc7867f DW	255	resource_size(res));
9049771f DW	256	if (error)
	257	goto err_pfn_remap;
	258
f931ab47	259	mem_hotplug_begin();
69324b8f DW	260
	261	/*
	262	* For device private memory we call add_pages() as we only need to
	263	* allocate and initialize struct page for the device memory. More-
	264	* over the device memory is un-accessible thus we do not want to
	265	* create a linear mapping for the memory like arch_add_memory()
	266	* would do.
	267	*
	268	* For all other device memory types, which are accessible by
	269	* the CPU, we do want the linear mapping and thus use
	270	* arch_add_memory().
	271	*/
	272	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
7cc7867f DW	273	error = add_pages(nid, PHYS_PFN(res->start),
7cc7867f DW	274	PHYS_PFN(resource_size(res)), &restrictions);
69324b8f	275	} else {
7cc7867f	276	error = kasan_add_zero_shadow(__va(res->start), resource_size(res));
69324b8f DW	277	if (error) {
	278	mem_hotplug_done();
	279	goto err_kasan;
	280	}
	281
7cc7867f	282	error = arch_add_memory(nid, res->start, resource_size(res),
940519f0	283	&restrictions);
69324b8f DW	284	}
	285
	286	if (!error) {
	287	struct zone *zone;
	288
	289	zone = &NODE_DATA(nid)->node_zones[ZONE_DEVICE];
7cc7867f DW	290	move_pfn_range_to_zone(zone, PHYS_PFN(res->start),
7cc7867f DW	291	PHYS_PFN(resource_size(res)), restrictions.altmap);
0207df4f AR	292	}
0207df4f AR	293
f931ab47	294	mem_hotplug_done();
9476df7d DW	295	if (error)
9476df7d DW	296	goto err_add_memory;
41e94a85	297
966cf44f AD	298	/*
	299	* Initialization of the pages has been deferred until now in order
	300	* to allow us to do the work while not holding the hotplug lock.
	301	*/
	302	memmap_init_zone_device(&NODE_DATA(nid)->node_zones[ZONE_DEVICE],
7cc7867f DW	303	PHYS_PFN(res->start),
7cc7867f DW	304	PHYS_PFN(resource_size(res)), pgmap);
966cf44f	305	percpu_ref_get_many(pgmap->ref, pfn_end(pgmap) - pfn_first(pgmap));
41e94a85	306	return __va(res->start);
9476df7d DW	307
9476df7d DW	308	err_add_memory:
7cc7867f	309	kasan_remove_zero_shadow(__va(res->start), resource_size(res));
0207df4f	310	err_kasan:
7cc7867f	311	untrack_pfn(NULL, PHYS_PFN(res->start), resource_size(res));
9049771f	312	err_pfn_remap:
bcfa4b72 MW	313	pgmap_array_delete(res);
bcfa4b72 MW	314	err_array:
24917f6b CH	315	dev_pagemap_kill(pgmap);
24917f6b CH	316	dev_pagemap_cleanup(pgmap);
6f42193f	317	devmap_managed_enable_put();
9476df7d	318	return ERR_PTR(error);
41e94a85	319	}
6869b7b2 CH	320	EXPORT_SYMBOL_GPL(memremap_pages);
	321
	322	/**
	323	* devm_memremap_pages - remap and provide memmap backing for the given resource
	324	* @dev: hosting device for @res
	325	* @pgmap: pointer to a struct dev_pagemap
	326	*
	327	* Notes:
	328	* 1/ At a minimum the res and type members of @pgmap must be initialized
	329	* by the caller before passing it to this function
	330	*
	331	* 2/ The altmap field may optionally be initialized, in which case
	332	* PGMAP_ALTMAP_VALID must be set in pgmap->flags.
	333	*
	334	* 3/ The ref field may optionally be provided, in which pgmap->ref must be
	335	* 'live' on entry and will be killed and reaped at
	336	* devm_memremap_pages_release() time, or if this routine fails.
	337	*
	338	* 4/ res is expected to be a host memory range that could feasibly be
	339	* treated as a "System RAM" range, i.e. not a device mmio range, but
	340	* this is not enforced.
	341	*/
	342	void devm_memremap_pages(struct device dev, struct dev_pagemap *pgmap)
	343	{
	344	int error;
	345	void *ret;
	346
	347	ret = memremap_pages(pgmap, dev_to_node(dev));
	348	if (IS_ERR(ret))
	349	return ret;
	350
	351	error = devm_add_action_or_reset(dev, devm_memremap_pages_release,
	352	pgmap);
	353	if (error)
	354	return ERR_PTR(error);
	355	return ret;
	356	}
808153e1	357	EXPORT_SYMBOL_GPL(devm_memremap_pages);
4b94ffdc	358
2e3f139e DW	359	void devm_memunmap_pages(struct device dev, struct dev_pagemap pgmap)
	360	{
	361	devm_release_action(dev, devm_memremap_pages_release, pgmap);
	362	}
	363	EXPORT_SYMBOL_GPL(devm_memunmap_pages);
	364
4b94ffdc DW	365	unsigned long vmem_altmap_offset(struct vmem_altmap *altmap)
	366	{
	367	/* number of pfns from base where pfn_to_page() is valid */
514caf23 CH	368	if (altmap)
	369	return altmap->reserve + altmap->free;
	370	return 0;
4b94ffdc DW	371	}
	372
	373	void vmem_altmap_free(struct vmem_altmap *altmap, unsigned long nr_pfns)
	374	{
	375	altmap->alloc -= nr_pfns;
	376	}
	377
0822acb8 CH	378	/**
	379	* get_dev_pagemap() - take a new live reference on the dev_pagemap for @pfn
	380	* @pfn: page frame number to lookup page_map
	381	* @pgmap: optional known pgmap that already has a reference
	382	*
832d7aa0 CH	383	* If @pgmap is non-NULL and covers @pfn it will be returned as-is. If @pgmap
832d7aa0 CH	384	* is non-NULL but does not cover @pfn the reference to it will be released.
0822acb8 CH	385	*/
	386	struct dev_pagemap *get_dev_pagemap(unsigned long pfn,
	387	struct dev_pagemap *pgmap)
	388	{
0822acb8 CH	389	resource_size_t phys = PFN_PHYS(pfn);
	390
	391	/*
832d7aa0	392	* In the cached case we're already holding a live reference.
0822acb8	393	*/
832d7aa0	394	if (pgmap) {
e7744aa2	395	if (phys >= pgmap->res.start && phys <= pgmap->res.end)
832d7aa0 CH	396	return pgmap;
832d7aa0 CH	397	put_dev_pagemap(pgmap);
0822acb8 CH	398	}
	399
	400	/* fall back to slow path lookup */
	401	rcu_read_lock();
bcfa4b72	402	pgmap = xa_load(&pgmap_array, PHYS_PFN(phys));
0822acb8 CH	403	if (pgmap && !percpu_ref_tryget_live(pgmap->ref))
	404	pgmap = NULL;
	405	rcu_read_unlock();
	406
	407	return pgmap;
	408	}
e7638488	409	EXPORT_SYMBOL_GPL(get_dev_pagemap);
7b2d55d2	410
e7638488	411	#ifdef CONFIG_DEV_PAGEMAP_OPS
e7638488	412	void __put_devmap_managed_page(struct page *page)
7b2d55d2 JG	413	{
	414	int count = page_ref_dec_return(page);
	415
	416	/*
	417	* If refcount is 1 then page is freed and refcount is stable as nobody
	418	* holds a reference on the page.
	419	*/
	420	if (count == 1) {
	421	/* Clear Active bit in case of parallel mark_page_accessed */
	422	__ClearPageActive(page);
	423	__ClearPageWaiters(page);
	424
c733a828	425	mem_cgroup_uncharge(page);
7b2d55d2	426
7ab0ad0e RC	427	/*
	428	* When a device_private page is freed, the page->mapping field
	429	* may still contain a (stale) mapping value. For example, the
	430	* lower bits of page->mapping may still identify the page as
	431	* an anonymous page. Ultimately, this entire field is just
	432	* stale and wrong, and it will cause errors if not cleared.
	433	* One example is:
	434	*
	435	* migrate_vma_pages()
	436	* migrate_vma_insert_page()
	437	* page_add_new_anon_rmap()
	438	* __page_set_anon_rmap()
	439	* ...checks page->mapping, via PageAnon(page) call,
	440	* and incorrectly concludes that the page is an
	441	* anonymous page. Therefore, it incorrectly,
	442	* silently fails to set up the new anon rmap.
	443	*
	444	* For other types of ZONE_DEVICE pages, migration is either
	445	* handled differently or not done at all, so there is no need
	446	* to clear page->mapping.
	447	*/
	448	if (is_device_private_page(page))
	449	page->mapping = NULL;
	450
80a72d0a	451	page->pgmap->ops->page_free(page);
7b2d55d2 JG	452	} else if (!count)
	453	__put_page(page);
	454	}
31c5bda3	455	EXPORT_SYMBOL(__put_devmap_managed_page);
e7638488	456	#endif /* CONFIG_DEV_PAGEMAP_OPS */