[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/mmzone.h>
#include <linux/swapops.h>
#include <linux/types.h>
#include <linux/wait_bit.h>
#include <linux/xarray.h>

static DEFINE_XARRAY(pgmap_array);

/*
 * The memremap() and memremap_pages() interfaces are alternately used
 * to map persistent memory namespaces. These interfaces place different
 * constraints on the alignment and size of the mapping (namespace).
 * memremap() can map individual PAGE_SIZE pages. memremap_pages() can
 * only map subsections (2MB), and at least one architecture (PowerPC)
 * the minimum mapping granularity of memremap_pages() is 16MB.
 *
 * The role of memremap_compat_align() is to communicate the minimum
 * arch supported alignment of a namespace such that it can freely
 * switch modes without violating the arch constraint. Namely, do not
 * allow a namespace to be PAGE_SIZE aligned since that namespace may be
 * reconfigured into a mode that requires SUBSECTION_SIZE alignment.
 */
#ifndef CONFIG_ARCH_HAS_MEMREMAP_COMPAT_ALIGN
unsigned long memremap_compat_align(void)
{
	return SUBSECTION_SIZE;
}
EXPORT_SYMBOL_GPL(memremap_compat_align);
#endif

#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 range *range)
{
	xa_store_range(&pgmap_array, PHYS_PFN(range->start), PHYS_PFN(range->end),
			NULL, GFP_KERNEL);
	synchronize_rcu();
}

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

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

	return (range->start + range_len(range)) >> 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 range *range = &pgmap->range;
	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(range->start),
				   PHYS_PFN(range_len(range)));
	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
		__remove_pages(PHYS_PFN(range->start),
			       PHYS_PFN(range_len(range)), NULL);
	} else {
		arch_remove_memory(nid, range->start, range_len(range),
				pgmap_altmap(pgmap));
		kasan_remove_zero_shadow(__va(range->start), range_len(range));
	}
	mem_hotplug_done();

	untrack_pfn(NULL, PHYS_PFN(range->start), range_len(range));
	pgmap_array_delete(range);
	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 range *range = &pgmap->range;
	struct dev_pagemap *conflict_pgmap;
	struct mhp_params params = {
		/*
		 * We do not want any optional features only our own memmap
		 */
		.altmap = pgmap_altmap(pgmap),
		.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_GENERIC:
		need_devmap_managed = false;
		break;
	case MEMORY_DEVICE_PCI_P2PDMA:
		params.pgprot = pgprot_noncached(params.pgprot);
		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(range->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(range->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(range->start, range_len(range),
		IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE);

	if (is_ram != REGION_DISJOINT) {
		WARN_ONCE(1, "attempted on %s region %#llx-%#llx\n",
				is_ram == REGION_MIXED ? "mixed" : "ram",
				range->start, range->end);
		error = -ENXIO;
		goto err_array;
	}

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

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

	error = track_pfn_remap(NULL, &params.pgprot, PHYS_PFN(range->start), 0,
			range_len(range));
	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(range->start),
				PHYS_PFN(range_len(range)), &params);
	} else {
		error = kasan_add_zero_shadow(__va(range->start), range_len(range));
		if (error) {
			mem_hotplug_done();
			goto err_kasan;
		}

		error = arch_add_memory(nid, range->start, range_len(range),
					&params);
	}

	if (!error) {
		struct zone *zone;

		zone = &NODE_DATA(nid)->node_zones[ZONE_DEVICE];
		move_pfn_range_to_zone(zone, PHYS_PFN(range->start),
				PHYS_PFN(range_len(range)), params.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(range->start),
				PHYS_PFN(range_len(range)), pgmap);
	percpu_ref_get_many(pgmap->ref, pfn_end(pgmap) - pfn_first(pgmap));
	return __va(range->start);

 err_add_memory:
	kasan_remove_zero_shadow(__va(range->start), range_len(range));
 err_kasan:
	untrack_pfn(NULL, PHYS_PFN(range->start), range_len(range));
 err_pfn_remap:
	pgmap_array_delete(range);
 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/ range 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->range.start && phys <= pgmap->range.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	9	#include <linux/swap.h>
9ffc1d19	10	#include <linux/mmzone.h>
5042db43	11	#include <linux/swapops.h>
bcfa4b72	12	#include <linux/types.h>
e7638488	13	#include <linux/wait_bit.h>
bcfa4b72	14	#include <linux/xarray.h>
92281dee	15
bcfa4b72	16	static DEFINE_XARRAY(pgmap_array);
9476df7d	17
9ffc1d19 DW	18	/*
	19	* The memremap() and memremap_pages() interfaces are alternately used
	20	* to map persistent memory namespaces. These interfaces place different
	21	* constraints on the alignment and size of the mapping (namespace).
	22	* memremap() can map individual PAGE_SIZE pages. memremap_pages() can
	23	* only map subsections (2MB), and at least one architecture (PowerPC)
	24	* the minimum mapping granularity of memremap_pages() is 16MB.
	25	*
	26	* The role of memremap_compat_align() is to communicate the minimum
	27	* arch supported alignment of a namespace such that it can freely
	28	* switch modes without violating the arch constraint. Namely, do not
	29	* allow a namespace to be PAGE_SIZE aligned since that namespace may be
	30	* reconfigured into a mode that requires SUBSECTION_SIZE alignment.
	31	*/
	32	#ifndef CONFIG_ARCH_HAS_MEMREMAP_COMPAT_ALIGN
	33	unsigned long memremap_compat_align(void)
	34	{
	35	return SUBSECTION_SIZE;
	36	}
	37	EXPORT_SYMBOL_GPL(memremap_compat_align);
	38	#endif
	39
f6a55e1a CH	40	#ifdef CONFIG_DEV_PAGEMAP_OPS
	41	DEFINE_STATIC_KEY_FALSE(devmap_managed_key);
	42	EXPORT_SYMBOL(devmap_managed_key);
	43	static atomic_t devmap_managed_enable;
	44
6f42193f	45	static void devmap_managed_enable_put(void)
f6a55e1a CH	46	{
	47	if (atomic_dec_and_test(&devmap_managed_enable))
	48	static_branch_disable(&devmap_managed_key);
	49	}
	50
6f42193f	51	static int devmap_managed_enable_get(struct dev_pagemap *pgmap)
f6a55e1a	52	{
429589d6 DW	53	if (pgmap->type == MEMORY_DEVICE_PRIVATE &&
429589d6 DW	54	(!pgmap->ops \|\| !pgmap->ops->page_free)) {
f6a55e1a CH	55	WARN(1, "Missing page_free method\n");
	56	return -EINVAL;
	57	}
	58
	59	if (atomic_inc_return(&devmap_managed_enable) == 1)
	60	static_branch_enable(&devmap_managed_key);
6f42193f	61	return 0;
f6a55e1a CH	62	}
f6a55e1a CH	63	#else
6f42193f	64	static int devmap_managed_enable_get(struct dev_pagemap *pgmap)
f6a55e1a CH	65	{
	66	return -EINVAL;
	67	}
6f42193f CH	68	static void devmap_managed_enable_put(void)
	69	{
	70	}
f6a55e1a CH	71	#endif /* CONFIG_DEV_PAGEMAP_OPS */
f6a55e1a CH	72
a4574f63	73	static void pgmap_array_delete(struct range *range)
ab1b597e	74	{
a4574f63	75	xa_store_range(&pgmap_array, PHYS_PFN(range->start), PHYS_PFN(range->end),
bcfa4b72	76	NULL, GFP_KERNEL);
ab1b597e	77	synchronize_rcu();
9476df7d DW	78	}
9476df7d DW	79
e7744aa2	80	static unsigned long pfn_first(struct dev_pagemap *pgmap)
5c2c2587	81	{
a4574f63	82	return PHYS_PFN(pgmap->range.start) +
514caf23	83	vmem_altmap_offset(pgmap_altmap(pgmap));
5c2c2587 DW	84	}
5c2c2587 DW	85
e7744aa2	86	static unsigned long pfn_end(struct dev_pagemap *pgmap)
5c2c2587	87	{
a4574f63	88	const struct range *range = &pgmap->range;
5c2c2587	89
a4574f63	90	return (range->start + range_len(range)) >> PAGE_SHIFT;
5c2c2587 DW	91	}
5c2c2587 DW	92
949b9325 DW	93	static unsigned long pfn_next(unsigned long pfn)
	94	{
	95	if (pfn % 1024 == 0)
	96	cond_resched();
	97	return pfn + 1;
	98	}
	99
5c2c2587	100	#define for_each_device_pfn(pfn, map) \
949b9325	101	for (pfn = pfn_first(map); pfn < pfn_end(map); pfn = pfn_next(pfn))
5c2c2587	102
24917f6b CH	103	static void dev_pagemap_kill(struct dev_pagemap *pgmap)
	104	{
	105	if (pgmap->ops && pgmap->ops->kill)
	106	pgmap->ops->kill(pgmap);
	107	else
	108	percpu_ref_kill(pgmap->ref);
	109	}
	110
	111	static void dev_pagemap_cleanup(struct dev_pagemap *pgmap)
	112	{
	113	if (pgmap->ops && pgmap->ops->cleanup) {
	114	pgmap->ops->cleanup(pgmap);
	115	} else {
	116	wait_for_completion(&pgmap->done);
	117	percpu_ref_exit(pgmap->ref);
	118	}
06282373 DW	119	/*
	120	* Undo the pgmap ref assignment for the internal case as the
	121	* caller may re-enable the same pgmap.
	122	*/
	123	if (pgmap->ref == &pgmap->internal_ref)
	124	pgmap->ref = NULL;
24917f6b CH	125	}
24917f6b CH	126
6869b7b2	127	void memunmap_pages(struct dev_pagemap *pgmap)
41e94a85	128	{
a4574f63	129	struct range *range = &pgmap->range;
77e080e7	130	struct page *first_page;
71389703	131	unsigned long pfn;
2c2a5af6	132	int nid;
71389703	133
24917f6b	134	dev_pagemap_kill(pgmap);
e7744aa2	135	for_each_device_pfn(pfn, pgmap)
71389703	136	put_page(pfn_to_page(pfn));
24917f6b	137	dev_pagemap_cleanup(pgmap);
9476df7d	138
77e080e7 AK	139	/* make sure to access a memmap that was actually initialized */
	140	first_page = pfn_to_page(pfn_first(pgmap));
	141
41e94a85	142	/* pages are dead and unused, undo the arch mapping */
77e080e7	143	nid = page_to_nid(first_page);
2c2a5af6	144
f931ab47	145	mem_hotplug_begin();
a4574f63 DW	146	remove_pfn_range_from_zone(page_zone(first_page), PHYS_PFN(range->start),
a4574f63 DW	147	PHYS_PFN(range_len(range)));
69324b8f	148	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
a4574f63 DW	149	__remove_pages(PHYS_PFN(range->start),
a4574f63 DW	150	PHYS_PFN(range_len(range)), NULL);
69324b8f	151	} else {
a4574f63	152	arch_remove_memory(nid, range->start, range_len(range),
514caf23	153	pgmap_altmap(pgmap));
a4574f63	154	kasan_remove_zero_shadow(__va(range->start), range_len(range));
69324b8f	155	}
f931ab47	156	mem_hotplug_done();
b5d24fda	157
a4574f63 DW	158	untrack_pfn(NULL, PHYS_PFN(range->start), range_len(range));
a4574f63 DW	159	pgmap_array_delete(range);
fdc029b1	160	WARN_ONCE(pgmap->altmap.alloc, "failed to free all reserved pages\n");
6f42193f	161	devmap_managed_enable_put();
9476df7d	162	}
6869b7b2 CH	163	EXPORT_SYMBOL_GPL(memunmap_pages);
	164
	165	static void devm_memremap_pages_release(void *data)
	166	{
	167	memunmap_pages(data);
	168	}
9476df7d	169
24917f6b CH	170	static void dev_pagemap_percpu_release(struct percpu_ref *ref)
	171	{
	172	struct dev_pagemap *pgmap =
	173	container_of(ref, struct dev_pagemap, internal_ref);
	174
	175	complete(&pgmap->done);
	176	}
	177
6869b7b2 CH	178	/*
	179	* Not device managed version of dev_memremap_pages, undone by
	180	* memunmap_pages(). Please use dev_memremap_pages if you have a struct
	181	* device available.
4b94ffdc	182	*/
6869b7b2	183	void memremap_pages(struct dev_pagemap pgmap, int nid)
41e94a85	184	{
a4574f63	185	struct range *range = &pgmap->range;
966cf44f	186	struct dev_pagemap *conflict_pgmap;
f5637d3b	187	struct mhp_params params = {
940519f0 MH	188	/*
940519f0 MH	189	* We do not want any optional features only our own memmap
7cc7867f	190	*/
514caf23	191	.altmap = pgmap_altmap(pgmap),
bfeb022f	192	.pgprot = PAGE_KERNEL,
940519f0	193	};
6869b7b2	194	int error, is_ram;
f6a55e1a	195	bool need_devmap_managed = true;
5f29a77c	196
3ed2dcdf CH	197	switch (pgmap->type) {
	198	case MEMORY_DEVICE_PRIVATE:
	199	if (!IS_ENABLED(CONFIG_DEVICE_PRIVATE)) {
	200	WARN(1, "Device private memory not supported\n");
	201	return ERR_PTR(-EINVAL);
	202	}
897e6365 CH	203	if (!pgmap->ops \|\| !pgmap->ops->migrate_to_ram) {
	204	WARN(1, "Missing migrate_to_ram method\n");
	205	return ERR_PTR(-EINVAL);
	206	}
f894ddd5 CH	207	if (!pgmap->owner) {
	208	WARN(1, "Missing owner\n");
	209	return ERR_PTR(-EINVAL);
	210	}
3ed2dcdf CH	211	break;
	212	case MEMORY_DEVICE_FS_DAX:
	213	if (!IS_ENABLED(CONFIG_ZONE_DEVICE) \|\|
	214	IS_ENABLED(CONFIG_FS_DAX_LIMITED)) {
	215	WARN(1, "File system DAX not supported\n");
	216	return ERR_PTR(-EINVAL);
	217	}
	218	break;
4533d3ae	219	case MEMORY_DEVICE_GENERIC:
a50d8d98 LG	220	need_devmap_managed = false;
a50d8d98 LG	221	break;
3ed2dcdf	222	case MEMORY_DEVICE_PCI_P2PDMA:
a50d8d98	223	params.pgprot = pgprot_noncached(params.pgprot);
f6a55e1a	224	need_devmap_managed = false;
3ed2dcdf CH	225	break;
	226	default:
	227	WARN(1, "Invalid pgmap type %d\n", pgmap->type);
	228	break;
	229	}
	230
24917f6b CH	231	if (!pgmap->ref) {
	232	if (pgmap->ops && (pgmap->ops->kill \|\| pgmap->ops->cleanup))
	233	return ERR_PTR(-EINVAL);
	234
	235	init_completion(&pgmap->done);
	236	error = percpu_ref_init(&pgmap->internal_ref,
	237	dev_pagemap_percpu_release, 0, GFP_KERNEL);
	238	if (error)
	239	return ERR_PTR(error);
	240	pgmap->ref = &pgmap->internal_ref;
	241	} else {
	242	if (!pgmap->ops \|\| !pgmap->ops->kill \|\| !pgmap->ops->cleanup) {
	243	WARN(1, "Missing reference count teardown definition\n");
	244	return ERR_PTR(-EINVAL);
	245	}
50f44ee7	246	}
a95c90f1	247
f6a55e1a	248	if (need_devmap_managed) {
6f42193f	249	error = devmap_managed_enable_get(pgmap);
f6a55e1a CH	250	if (error)
	251	return ERR_PTR(error);
	252	}
	253
a4574f63	254	conflict_pgmap = get_dev_pagemap(PHYS_PFN(range->start), NULL);
15d36fec	255	if (conflict_pgmap) {
6869b7b2	256	WARN(1, "Conflicting mapping in same section\n");
15d36fec	257	put_dev_pagemap(conflict_pgmap);
50f44ee7 DW	258	error = -ENOMEM;
50f44ee7 DW	259	goto err_array;
15d36fec DJ	260	}
15d36fec DJ	261
a4574f63	262	conflict_pgmap = get_dev_pagemap(PHYS_PFN(range->end), NULL);
15d36fec	263	if (conflict_pgmap) {
6869b7b2	264	WARN(1, "Conflicting mapping in same section\n");
15d36fec	265	put_dev_pagemap(conflict_pgmap);
50f44ee7 DW	266	error = -ENOMEM;
50f44ee7 DW	267	goto err_array;
15d36fec DJ	268	}
15d36fec DJ	269
a4574f63	270	is_ram = region_intersects(range->start, range_len(range),
d37a14bb	271	IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE);
41e94a85	272
06489cfb	273	if (is_ram != REGION_DISJOINT) {
a4574f63 DW	274	WARN_ONCE(1, "attempted on %s region %#llx-%#llx\n",
	275	is_ram == REGION_MIXED ? "mixed" : "ram",
	276	range->start, range->end);
a95c90f1 DW	277	error = -ENXIO;
a95c90f1 DW	278	goto err_array;
41e94a85 CH	279	}
41e94a85 CH	280
a4574f63 DW	281	error = xa_err(xa_store_range(&pgmap_array, PHYS_PFN(range->start),
a4574f63 DW	282	PHYS_PFN(range->end), pgmap, GFP_KERNEL));
9476df7d	283	if (error)
bcfa4b72	284	goto err_array;
9476df7d	285
41e94a85	286	if (nid < 0)
7eff93b7	287	nid = numa_mem_id();
41e94a85	288
a4574f63 DW	289	error = track_pfn_remap(NULL, &params.pgprot, PHYS_PFN(range->start), 0,
a4574f63 DW	290	range_len(range));
9049771f DW	291	if (error)
	292	goto err_pfn_remap;
	293
f931ab47	294	mem_hotplug_begin();
69324b8f DW	295
	296	/*
	297	* For device private memory we call add_pages() as we only need to
	298	* allocate and initialize struct page for the device memory. More-
	299	* over the device memory is un-accessible thus we do not want to
	300	* create a linear mapping for the memory like arch_add_memory()
	301	* would do.
	302	*
	303	* For all other device memory types, which are accessible by
	304	* the CPU, we do want the linear mapping and thus use
	305	* arch_add_memory().
	306	*/
	307	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
a4574f63 DW	308	error = add_pages(nid, PHYS_PFN(range->start),
a4574f63 DW	309	PHYS_PFN(range_len(range)), &params);
69324b8f	310	} else {
a4574f63	311	error = kasan_add_zero_shadow(__va(range->start), range_len(range));
69324b8f DW	312	if (error) {
	313	mem_hotplug_done();
	314	goto err_kasan;
	315	}
	316
a4574f63	317	error = arch_add_memory(nid, range->start, range_len(range),
f5637d3b	318	&params);
69324b8f DW	319	}
	320
	321	if (!error) {
	322	struct zone *zone;
	323
	324	zone = &NODE_DATA(nid)->node_zones[ZONE_DEVICE];
a4574f63 DW	325	move_pfn_range_to_zone(zone, PHYS_PFN(range->start),
a4574f63 DW	326	PHYS_PFN(range_len(range)), params.altmap);
0207df4f AR	327	}
0207df4f AR	328
f931ab47	329	mem_hotplug_done();
9476df7d DW	330	if (error)
9476df7d DW	331	goto err_add_memory;
41e94a85	332
966cf44f AD	333	/*
	334	* Initialization of the pages has been deferred until now in order
	335	* to allow us to do the work while not holding the hotplug lock.
	336	*/
	337	memmap_init_zone_device(&NODE_DATA(nid)->node_zones[ZONE_DEVICE],
a4574f63 DW	338	PHYS_PFN(range->start),
a4574f63 DW	339	PHYS_PFN(range_len(range)), pgmap);
966cf44f	340	percpu_ref_get_many(pgmap->ref, pfn_end(pgmap) - pfn_first(pgmap));
a4574f63	341	return __va(range->start);
9476df7d DW	342
9476df7d DW	343	err_add_memory:
a4574f63	344	kasan_remove_zero_shadow(__va(range->start), range_len(range));
0207df4f	345	err_kasan:
a4574f63	346	untrack_pfn(NULL, PHYS_PFN(range->start), range_len(range));
9049771f	347	err_pfn_remap:
a4574f63	348	pgmap_array_delete(range);
bcfa4b72	349	err_array:
24917f6b CH	350	dev_pagemap_kill(pgmap);
24917f6b CH	351	dev_pagemap_cleanup(pgmap);
6f42193f	352	devmap_managed_enable_put();
9476df7d	353	return ERR_PTR(error);
41e94a85	354	}
6869b7b2 CH	355	EXPORT_SYMBOL_GPL(memremap_pages);
	356
	357	/**
	358	* devm_memremap_pages - remap and provide memmap backing for the given resource
	359	* @dev: hosting device for @res
	360	* @pgmap: pointer to a struct dev_pagemap
	361	*
	362	* Notes:
	363	* 1/ At a minimum the res and type members of @pgmap must be initialized
	364	* by the caller before passing it to this function
	365	*
	366	* 2/ The altmap field may optionally be initialized, in which case
	367	* PGMAP_ALTMAP_VALID must be set in pgmap->flags.
	368	*
	369	* 3/ The ref field may optionally be provided, in which pgmap->ref must be
	370	* 'live' on entry and will be killed and reaped at
	371	* devm_memremap_pages_release() time, or if this routine fails.
	372	*
a4574f63	373	* 4/ range is expected to be a host memory range that could feasibly be
6869b7b2 CH	374	* treated as a "System RAM" range, i.e. not a device mmio range, but
	375	* this is not enforced.
	376	*/
	377	void devm_memremap_pages(struct device dev, struct dev_pagemap *pgmap)
	378	{
	379	int error;
	380	void *ret;
	381
	382	ret = memremap_pages(pgmap, dev_to_node(dev));
	383	if (IS_ERR(ret))
	384	return ret;
	385
	386	error = devm_add_action_or_reset(dev, devm_memremap_pages_release,
	387	pgmap);
	388	if (error)
	389	return ERR_PTR(error);
	390	return ret;
	391	}
808153e1	392	EXPORT_SYMBOL_GPL(devm_memremap_pages);
4b94ffdc	393
2e3f139e DW	394	void devm_memunmap_pages(struct device dev, struct dev_pagemap pgmap)
	395	{
	396	devm_release_action(dev, devm_memremap_pages_release, pgmap);
	397	}
	398	EXPORT_SYMBOL_GPL(devm_memunmap_pages);
	399
4b94ffdc DW	400	unsigned long vmem_altmap_offset(struct vmem_altmap *altmap)
	401	{
	402	/* number of pfns from base where pfn_to_page() is valid */
514caf23 CH	403	if (altmap)
	404	return altmap->reserve + altmap->free;
	405	return 0;
4b94ffdc DW	406	}
	407
	408	void vmem_altmap_free(struct vmem_altmap *altmap, unsigned long nr_pfns)
	409	{
	410	altmap->alloc -= nr_pfns;
	411	}
	412
0822acb8 CH	413	/**
	414	* get_dev_pagemap() - take a new live reference on the dev_pagemap for @pfn
	415	* @pfn: page frame number to lookup page_map
	416	* @pgmap: optional known pgmap that already has a reference
	417	*
832d7aa0 CH	418	* If @pgmap is non-NULL and covers @pfn it will be returned as-is. If @pgmap
832d7aa0 CH	419	* is non-NULL but does not cover @pfn the reference to it will be released.
0822acb8 CH	420	*/
	421	struct dev_pagemap *get_dev_pagemap(unsigned long pfn,
	422	struct dev_pagemap *pgmap)
	423	{
0822acb8 CH	424	resource_size_t phys = PFN_PHYS(pfn);
	425
	426	/*
832d7aa0	427	* In the cached case we're already holding a live reference.
0822acb8	428	*/
832d7aa0	429	if (pgmap) {
a4574f63	430	if (phys >= pgmap->range.start && phys <= pgmap->range.end)
832d7aa0 CH	431	return pgmap;
832d7aa0 CH	432	put_dev_pagemap(pgmap);
0822acb8 CH	433	}
	434
	435	/* fall back to slow path lookup */
	436	rcu_read_lock();
bcfa4b72	437	pgmap = xa_load(&pgmap_array, PHYS_PFN(phys));
0822acb8 CH	438	if (pgmap && !percpu_ref_tryget_live(pgmap->ref))
	439	pgmap = NULL;
	440	rcu_read_unlock();
	441
	442	return pgmap;
	443	}
e7638488	444	EXPORT_SYMBOL_GPL(get_dev_pagemap);
7b2d55d2	445
e7638488	446	#ifdef CONFIG_DEV_PAGEMAP_OPS
07d80269	447	void free_devmap_managed_page(struct page *page)
7b2d55d2	448	{
429589d6 DW	449	/* notify page idle for dax */
	450	if (!is_device_private_page(page)) {
	451	wake_up_var(&page->_refcount);
	452	return;
	453	}
7ab0ad0e	454
429589d6 DW	455	/* Clear Active bit in case of parallel mark_page_accessed */
	456	__ClearPageActive(page);
	457	__ClearPageWaiters(page);
	458
	459	mem_cgroup_uncharge(page);
	460
	461	/*
	462	* When a device_private page is freed, the page->mapping field
	463	* may still contain a (stale) mapping value. For example, the
	464	* lower bits of page->mapping may still identify the page as an
	465	* anonymous page. Ultimately, this entire field is just stale
	466	* and wrong, and it will cause errors if not cleared. One
	467	* example is:
	468	*
	469	* migrate_vma_pages()
	470	* migrate_vma_insert_page()
	471	* page_add_new_anon_rmap()
	472	* __page_set_anon_rmap()
	473	* ...checks page->mapping, via PageAnon(page) call,
	474	* and incorrectly concludes that the page is an
	475	* anonymous page. Therefore, it incorrectly,
	476	* silently fails to set up the new anon rmap.
	477	*
	478	* For other types of ZONE_DEVICE pages, migration is either
	479	* handled differently or not done at all, so there is no need
	480	* to clear page->mapping.
	481	*/
	482	page->mapping = NULL;
	483	page->pgmap->ops->page_free(page);
7b2d55d2	484	}
e7638488	485	#endif /* CONFIG_DEV_PAGEMAP_OPS */