[linux-2.6-block.git] / kernel / 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 *data)
{
	if (atomic_dec_and_test(&devmap_managed_enable))
		static_branch_disable(&devmap_managed_key);
}

static int devmap_managed_enable_get(struct device *dev, 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 devm_add_action_or_reset(dev, devmap_managed_enable_put, NULL);
}
#else
static int devmap_managed_enable_get(struct device *dev, struct dev_pagemap *pgmap)
{
	return -EINVAL;
}
#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 (pgmap->res.start >> PAGE_SHIFT) +
		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);
	}
}

static void devm_memremap_pages_release(void *data)
{
	struct dev_pagemap *pgmap = data;
	struct device *dev = pgmap->dev;
	struct resource *res = &pgmap->res;
	resource_size_t align_start, align_size;
	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 */
	align_start = res->start & ~(SECTION_SIZE - 1);
	align_size = ALIGN(res->start + resource_size(res), SECTION_SIZE)
		- align_start;

	nid = page_to_nid(pfn_to_page(align_start >> PAGE_SHIFT));

	mem_hotplug_begin();
	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
		pfn = align_start >> PAGE_SHIFT;
		__remove_pages(page_zone(pfn_to_page(pfn)), pfn,
				align_size >> PAGE_SHIFT, NULL);
	} else {
		arch_remove_memory(nid, align_start, align_size,
				pgmap_altmap(pgmap));
		kasan_remove_zero_shadow(__va(align_start), align_size);
	}
	mem_hotplug_done();

	untrack_pfn(NULL, PHYS_PFN(align_start), align_size);
	pgmap_array_delete(res);
	dev_WARN_ONCE(dev, pgmap->altmap.alloc,
		      "%s: failed to free all reserved pages\n", __func__);
}

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);
}

/**
 * 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)
{
	resource_size_t align_start, align_size, align_end;
	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, nid, 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(dev, pgmap);
		if (error)
			return ERR_PTR(error);
	}

	align_start = res->start & ~(SECTION_SIZE - 1);
	align_size = ALIGN(res->start + resource_size(res), SECTION_SIZE)
		- align_start;
	align_end = align_start + align_size - 1;

	conflict_pgmap = get_dev_pagemap(PHYS_PFN(align_start), NULL);
	if (conflict_pgmap) {
		dev_WARN(dev, "Conflicting mapping in same section\n");
		put_dev_pagemap(conflict_pgmap);
		error = -ENOMEM;
		goto err_array;
	}

	conflict_pgmap = get_dev_pagemap(PHYS_PFN(align_end), NULL);
	if (conflict_pgmap) {
		dev_WARN(dev, "Conflicting mapping in same section\n");
		put_dev_pagemap(conflict_pgmap);
		error = -ENOMEM;
		goto err_array;
	}

	is_ram = region_intersects(align_start, align_size,
		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;
	}

	pgmap->dev = dev;

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

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

	error = track_pfn_remap(NULL, &pgprot, PHYS_PFN(align_start), 0,
			align_size);
	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, align_start >> PAGE_SHIFT,
				align_size >> PAGE_SHIFT, &restrictions);
	} else {
		error = kasan_add_zero_shadow(__va(align_start), align_size);
		if (error) {
			mem_hotplug_done();
			goto err_kasan;
		}

		error = arch_add_memory(nid, align_start, align_size,
					&restrictions);
	}

	if (!error) {
		struct zone *zone;

		zone = &NODE_DATA(nid)->node_zones[ZONE_DEVICE];
		move_pfn_range_to_zone(zone, align_start >> PAGE_SHIFT,
				align_size >> PAGE_SHIFT, pgmap_altmap(pgmap));
	}

	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],
				align_start >> PAGE_SHIFT,
				align_size >> PAGE_SHIFT, pgmap);
	percpu_ref_get_many(pgmap->ref, pfn_end(pgmap) - pfn_first(pgmap));

	error = devm_add_action_or_reset(dev, devm_memremap_pages_release,
			pgmap);
	if (error)
		return ERR_PTR(error);

	return __va(res->start);

 err_add_memory:
	kasan_remove_zero_shadow(__va(align_start), align_size);
 err_kasan:
	untrack_pfn(NULL, PHYS_PFN(align_start), align_size);
 err_pfn_remap:
	pgmap_array_delete(res);
 err_array:
	dev_pagemap_kill(pgmap);
	dev_pagemap_cleanup(pgmap);
	return ERR_PTR(error);
}
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);

		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
	24	static void devmap_managed_enable_put(void *data)
	25	{
	26	if (atomic_dec_and_test(&devmap_managed_enable))
	27	static_branch_disable(&devmap_managed_key);
	28	}
	29
	30	static int devmap_managed_enable_get(struct device dev, struct dev_pagemap pgmap)
	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);
	39	return devm_add_action_or_reset(dev, devmap_managed_enable_put, NULL);
	40	}
	41	#else
	42	static int devmap_managed_enable_get(struct device dev, struct dev_pagemap pgmap)
	43	{
	44	return -EINVAL;
	45	}
	46	#endif /* CONFIG_DEV_PAGEMAP_OPS */
	47
bcfa4b72	48	static void pgmap_array_delete(struct resource *res)
ab1b597e	49	{
bcfa4b72 MW	50	xa_store_range(&pgmap_array, PHYS_PFN(res->start), PHYS_PFN(res->end),
bcfa4b72 MW	51	NULL, GFP_KERNEL);
ab1b597e	52	synchronize_rcu();
9476df7d DW	53	}
9476df7d DW	54
e7744aa2	55	static unsigned long pfn_first(struct dev_pagemap *pgmap)
5c2c2587	56	{
514caf23 CH	57	return (pgmap->res.start >> PAGE_SHIFT) +
514caf23 CH	58	vmem_altmap_offset(pgmap_altmap(pgmap));
5c2c2587 DW	59	}
5c2c2587 DW	60
e7744aa2	61	static unsigned long pfn_end(struct dev_pagemap *pgmap)
5c2c2587	62	{
e7744aa2	63	const struct resource *res = &pgmap->res;
5c2c2587 DW	64
	65	return (res->start + resource_size(res)) >> PAGE_SHIFT;
	66	}
	67
949b9325 DW	68	static unsigned long pfn_next(unsigned long pfn)
	69	{
	70	if (pfn % 1024 == 0)
	71	cond_resched();
	72	return pfn + 1;
	73	}
	74
5c2c2587	75	#define for_each_device_pfn(pfn, map) \
949b9325	76	for (pfn = pfn_first(map); pfn < pfn_end(map); pfn = pfn_next(pfn))
5c2c2587	77
24917f6b CH	78	static void dev_pagemap_kill(struct dev_pagemap *pgmap)
	79	{
	80	if (pgmap->ops && pgmap->ops->kill)
	81	pgmap->ops->kill(pgmap);
	82	else
	83	percpu_ref_kill(pgmap->ref);
	84	}
	85
	86	static void dev_pagemap_cleanup(struct dev_pagemap *pgmap)
	87	{
	88	if (pgmap->ops && pgmap->ops->cleanup) {
	89	pgmap->ops->cleanup(pgmap);
	90	} else {
	91	wait_for_completion(&pgmap->done);
	92	percpu_ref_exit(pgmap->ref);
	93	}
	94	}
	95
e8d51348	96	static void devm_memremap_pages_release(void *data)
41e94a85	97	{
e7744aa2	98	struct dev_pagemap *pgmap = data;
e8d51348	99	struct device *dev = pgmap->dev;
e7744aa2	100	struct resource *res = &pgmap->res;
9476df7d	101	resource_size_t align_start, align_size;
71389703	102	unsigned long pfn;
2c2a5af6	103	int nid;
71389703	104
24917f6b	105	dev_pagemap_kill(pgmap);
e7744aa2	106	for_each_device_pfn(pfn, pgmap)
71389703	107	put_page(pfn_to_page(pfn));
24917f6b	108	dev_pagemap_cleanup(pgmap);
9476df7d	109
41e94a85	110	/* pages are dead and unused, undo the arch mapping */
9476df7d	111	align_start = res->start & ~(SECTION_SIZE - 1);
10a0cd6e JS	112	align_size = ALIGN(res->start + resource_size(res), SECTION_SIZE)
10a0cd6e JS	113	- align_start;
b5d24fda	114
2c2a5af6 OS	115	nid = page_to_nid(pfn_to_page(align_start >> PAGE_SHIFT));
2c2a5af6 OS	116
f931ab47	117	mem_hotplug_begin();
69324b8f DW	118	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
	119	pfn = align_start >> PAGE_SHIFT;
	120	__remove_pages(page_zone(pfn_to_page(pfn)), pfn,
	121	align_size >> PAGE_SHIFT, NULL);
	122	} else {
2c2a5af6	123	arch_remove_memory(nid, align_start, align_size,
514caf23	124	pgmap_altmap(pgmap));
69324b8f DW	125	kasan_remove_zero_shadow(__va(align_start), align_size);
69324b8f DW	126	}
f931ab47	127	mem_hotplug_done();
b5d24fda	128
9049771f	129	untrack_pfn(NULL, PHYS_PFN(align_start), align_size);
bcfa4b72	130	pgmap_array_delete(res);
e7744aa2 LG	131	dev_WARN_ONCE(dev, pgmap->altmap.alloc,
e7744aa2 LG	132	"%s: failed to free all reserved pages\n", __func__);
9476df7d DW	133	}
9476df7d DW	134
24917f6b CH	135	static void dev_pagemap_percpu_release(struct percpu_ref *ref)
	136	{
	137	struct dev_pagemap *pgmap =
	138	container_of(ref, struct dev_pagemap, internal_ref);
	139
	140	complete(&pgmap->done);
	141	}
	142
4b94ffdc DW	143	/**
	144	* devm_memremap_pages - remap and provide memmap backing for the given resource
	145	* @dev: hosting device for @res
a95c90f1	146	* @pgmap: pointer to a struct dev_pagemap
4b94ffdc	147	*
5c2c2587	148	* Notes:
24917f6b CH	149	* 1/ At a minimum the res and type members of @pgmap must be initialized
24917f6b CH	150	* by the caller before passing it to this function
e8d51348	151	*
514caf23 CH	152	* 2/ The altmap field may optionally be initialized, in which case
514caf23 CH	153	* PGMAP_ALTMAP_VALID must be set in pgmap->flags.
e8d51348	154	*
24917f6b CH	155	* 3/ The ref field may optionally be provided, in which pgmap->ref must be
	156	* 'live' on entry and will be killed and reaped at
	157	* devm_memremap_pages_release() time, or if this routine fails.
5c2c2587	158	*
e8d51348	159	* 4/ res is expected to be a host memory range that could feasibly be
5c2c2587 DW	160	* treated as a "System RAM" range, i.e. not a device mmio range, but
5c2c2587 DW	161	* this is not enforced.
4b94ffdc	162	*/
e8d51348	163	void devm_memremap_pages(struct device dev, struct dev_pagemap *pgmap)
41e94a85	164	{
ab1b597e	165	resource_size_t align_start, align_size, align_end;
949b9325	166	struct resource *res = &pgmap->res;
966cf44f	167	struct dev_pagemap *conflict_pgmap;
940519f0 MH	168	struct mhp_restrictions restrictions = {
	169	/*
	170	* We do not want any optional features only our own memmap
	171	*/
514caf23	172	.altmap = pgmap_altmap(pgmap),
940519f0	173	};
9049771f	174	pgprot_t pgprot = PAGE_KERNEL;
949b9325	175	int error, nid, is_ram;
f6a55e1a	176	bool need_devmap_managed = true;
5f29a77c	177
3ed2dcdf CH	178	switch (pgmap->type) {
	179	case MEMORY_DEVICE_PRIVATE:
	180	if (!IS_ENABLED(CONFIG_DEVICE_PRIVATE)) {
	181	WARN(1, "Device private memory not supported\n");
	182	return ERR_PTR(-EINVAL);
	183	}
897e6365 CH	184	if (!pgmap->ops \|\| !pgmap->ops->migrate_to_ram) {
	185	WARN(1, "Missing migrate_to_ram method\n");
	186	return ERR_PTR(-EINVAL);
	187	}
3ed2dcdf CH	188	break;
	189	case MEMORY_DEVICE_FS_DAX:
	190	if (!IS_ENABLED(CONFIG_ZONE_DEVICE) \|\|
	191	IS_ENABLED(CONFIG_FS_DAX_LIMITED)) {
	192	WARN(1, "File system DAX not supported\n");
	193	return ERR_PTR(-EINVAL);
	194	}
	195	break;
	196	case MEMORY_DEVICE_DEVDAX:
	197	case MEMORY_DEVICE_PCI_P2PDMA:
f6a55e1a	198	need_devmap_managed = false;
3ed2dcdf CH	199	break;
	200	default:
	201	WARN(1, "Invalid pgmap type %d\n", pgmap->type);
	202	break;
	203	}
	204
24917f6b CH	205	if (!pgmap->ref) {
	206	if (pgmap->ops && (pgmap->ops->kill \|\| pgmap->ops->cleanup))
	207	return ERR_PTR(-EINVAL);
	208
	209	init_completion(&pgmap->done);
	210	error = percpu_ref_init(&pgmap->internal_ref,
	211	dev_pagemap_percpu_release, 0, GFP_KERNEL);
	212	if (error)
	213	return ERR_PTR(error);
	214	pgmap->ref = &pgmap->internal_ref;
	215	} else {
	216	if (!pgmap->ops \|\| !pgmap->ops->kill \|\| !pgmap->ops->cleanup) {
	217	WARN(1, "Missing reference count teardown definition\n");
	218	return ERR_PTR(-EINVAL);
	219	}
50f44ee7	220	}
a95c90f1	221
f6a55e1a CH	222	if (need_devmap_managed) {
	223	error = devmap_managed_enable_get(dev, pgmap);
	224	if (error)
	225	return ERR_PTR(error);
	226	}
	227
5f29a77c DW	228	align_start = res->start & ~(SECTION_SIZE - 1);
	229	align_size = ALIGN(res->start + resource_size(res), SECTION_SIZE)
	230	- align_start;
15d36fec DJ	231	align_end = align_start + align_size - 1;
	232
	233	conflict_pgmap = get_dev_pagemap(PHYS_PFN(align_start), NULL);
	234	if (conflict_pgmap) {
	235	dev_WARN(dev, "Conflicting mapping in same section\n");
	236	put_dev_pagemap(conflict_pgmap);
50f44ee7 DW	237	error = -ENOMEM;
50f44ee7 DW	238	goto err_array;
15d36fec DJ	239	}
	240
	241	conflict_pgmap = get_dev_pagemap(PHYS_PFN(align_end), NULL);
	242	if (conflict_pgmap) {
	243	dev_WARN(dev, "Conflicting mapping in same section\n");
	244	put_dev_pagemap(conflict_pgmap);
50f44ee7 DW	245	error = -ENOMEM;
50f44ee7 DW	246	goto err_array;
15d36fec DJ	247	}
15d36fec DJ	248
d37a14bb LT	249	is_ram = region_intersects(align_start, align_size,
d37a14bb LT	250	IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE);
41e94a85	251
06489cfb DW	252	if (is_ram != REGION_DISJOINT) {
	253	WARN_ONCE(1, "%s attempted on %s region %pr\n", __func__,
	254	is_ram == REGION_MIXED ? "mixed" : "ram", res);
a95c90f1 DW	255	error = -ENXIO;
a95c90f1 DW	256	goto err_array;
41e94a85 CH	257	}
41e94a85 CH	258
4b94ffdc	259	pgmap->dev = dev;
4b94ffdc	260
bcfa4b72 MW	261	error = xa_err(xa_store_range(&pgmap_array, PHYS_PFN(res->start),
bcfa4b72 MW	262	PHYS_PFN(res->end), pgmap, GFP_KERNEL));
9476df7d	263	if (error)
bcfa4b72	264	goto err_array;
9476df7d	265
41e94a85 CH	266	nid = dev_to_node(dev);
41e94a85 CH	267	if (nid < 0)
7eff93b7	268	nid = numa_mem_id();
41e94a85	269
9049771f DW	270	error = track_pfn_remap(NULL, &pgprot, PHYS_PFN(align_start), 0,
	271	align_size);
	272	if (error)
	273	goto err_pfn_remap;
	274
f931ab47	275	mem_hotplug_begin();
69324b8f DW	276
	277	/*
	278	* For device private memory we call add_pages() as we only need to
	279	* allocate and initialize struct page for the device memory. More-
	280	* over the device memory is un-accessible thus we do not want to
	281	* create a linear mapping for the memory like arch_add_memory()
	282	* would do.
	283	*
	284	* For all other device memory types, which are accessible by
	285	* the CPU, we do want the linear mapping and thus use
	286	* arch_add_memory().
	287	*/
	288	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
	289	error = add_pages(nid, align_start >> PAGE_SHIFT,
940519f0	290	align_size >> PAGE_SHIFT, &restrictions);
69324b8f DW	291	} else {
	292	error = kasan_add_zero_shadow(__va(align_start), align_size);
	293	if (error) {
	294	mem_hotplug_done();
	295	goto err_kasan;
	296	}
	297
940519f0 MH	298	error = arch_add_memory(nid, align_start, align_size,
940519f0 MH	299	&restrictions);
69324b8f DW	300	}
	301
	302	if (!error) {
	303	struct zone *zone;
	304
	305	zone = &NODE_DATA(nid)->node_zones[ZONE_DEVICE];
	306	move_pfn_range_to_zone(zone, align_start >> PAGE_SHIFT,
514caf23	307	align_size >> PAGE_SHIFT, pgmap_altmap(pgmap));
0207df4f AR	308	}
0207df4f AR	309
f931ab47	310	mem_hotplug_done();
9476df7d DW	311	if (error)
9476df7d DW	312	goto err_add_memory;
41e94a85	313
966cf44f AD	314	/*
	315	* Initialization of the pages has been deferred until now in order
	316	* to allow us to do the work while not holding the hotplug lock.
	317	*/
	318	memmap_init_zone_device(&NODE_DATA(nid)->node_zones[ZONE_DEVICE],
	319	align_start >> PAGE_SHIFT,
	320	align_size >> PAGE_SHIFT, pgmap);
	321	percpu_ref_get_many(pgmap->ref, pfn_end(pgmap) - pfn_first(pgmap));
e8d51348	322
a95c90f1 DW	323	error = devm_add_action_or_reset(dev, devm_memremap_pages_release,
	324	pgmap);
	325	if (error)
	326	return ERR_PTR(error);
e8d51348	327
41e94a85	328	return __va(res->start);
9476df7d DW	329
9476df7d DW	330	err_add_memory:
0207df4f AR	331	kasan_remove_zero_shadow(__va(align_start), align_size);
0207df4f AR	332	err_kasan:
9049771f DW	333	untrack_pfn(NULL, PHYS_PFN(align_start), align_size);
9049771f DW	334	err_pfn_remap:
bcfa4b72 MW	335	pgmap_array_delete(res);
bcfa4b72 MW	336	err_array:
24917f6b CH	337	dev_pagemap_kill(pgmap);
24917f6b CH	338	dev_pagemap_cleanup(pgmap);
9476df7d	339	return ERR_PTR(error);
41e94a85	340	}
808153e1	341	EXPORT_SYMBOL_GPL(devm_memremap_pages);
4b94ffdc	342
2e3f139e DW	343	void devm_memunmap_pages(struct device dev, struct dev_pagemap pgmap)
	344	{
	345	devm_release_action(dev, devm_memremap_pages_release, pgmap);
	346	}
	347	EXPORT_SYMBOL_GPL(devm_memunmap_pages);
	348
4b94ffdc DW	349	unsigned long vmem_altmap_offset(struct vmem_altmap *altmap)
	350	{
	351	/* number of pfns from base where pfn_to_page() is valid */
514caf23 CH	352	if (altmap)
	353	return altmap->reserve + altmap->free;
	354	return 0;
4b94ffdc DW	355	}
	356
	357	void vmem_altmap_free(struct vmem_altmap *altmap, unsigned long nr_pfns)
	358	{
	359	altmap->alloc -= nr_pfns;
	360	}
	361
0822acb8 CH	362	/**
	363	* get_dev_pagemap() - take a new live reference on the dev_pagemap for @pfn
	364	* @pfn: page frame number to lookup page_map
	365	* @pgmap: optional known pgmap that already has a reference
	366	*
832d7aa0 CH	367	* If @pgmap is non-NULL and covers @pfn it will be returned as-is. If @pgmap
832d7aa0 CH	368	* is non-NULL but does not cover @pfn the reference to it will be released.
0822acb8 CH	369	*/
	370	struct dev_pagemap *get_dev_pagemap(unsigned long pfn,
	371	struct dev_pagemap *pgmap)
	372	{
0822acb8 CH	373	resource_size_t phys = PFN_PHYS(pfn);
	374
	375	/*
832d7aa0	376	* In the cached case we're already holding a live reference.
0822acb8	377	*/
832d7aa0	378	if (pgmap) {
e7744aa2	379	if (phys >= pgmap->res.start && phys <= pgmap->res.end)
832d7aa0 CH	380	return pgmap;
832d7aa0 CH	381	put_dev_pagemap(pgmap);
0822acb8 CH	382	}
	383
	384	/* fall back to slow path lookup */
	385	rcu_read_lock();
bcfa4b72	386	pgmap = xa_load(&pgmap_array, PHYS_PFN(phys));
0822acb8 CH	387	if (pgmap && !percpu_ref_tryget_live(pgmap->ref))
	388	pgmap = NULL;
	389	rcu_read_unlock();
	390
	391	return pgmap;
	392	}
e7638488	393	EXPORT_SYMBOL_GPL(get_dev_pagemap);
7b2d55d2	394
e7638488	395	#ifdef CONFIG_DEV_PAGEMAP_OPS
e7638488	396	void __put_devmap_managed_page(struct page *page)
7b2d55d2 JG	397	{
	398	int count = page_ref_dec_return(page);
	399
	400	/*
	401	* If refcount is 1 then page is freed and refcount is stable as nobody
	402	* holds a reference on the page.
	403	*/
	404	if (count == 1) {
	405	/* Clear Active bit in case of parallel mark_page_accessed */
	406	__ClearPageActive(page);
	407	__ClearPageWaiters(page);
	408
c733a828	409	mem_cgroup_uncharge(page);
7b2d55d2	410
80a72d0a	411	page->pgmap->ops->page_free(page);
7b2d55d2 JG	412	} else if (!count)
	413	__put_page(page);
	414	}
31c5bda3	415	EXPORT_SYMBOL(__put_devmap_managed_page);
e7638488	416	#endif /* CONFIG_DEV_PAGEMAP_OPS */