[linux-block.git] / kernel / memremap.c

/* SPDX-License-Identifier: GPL-2.0 */
/* Copyright(c) 2015 Intel Corporation. All rights reserved. */
#include <linux/radix-tree.h>
#include <linux/device.h>
#include <linux/types.h>
#include <linux/pfn_t.h>
#include <linux/io.h>
#include <linux/kasan.h>
#include <linux/mm.h>
#include <linux/memory_hotplug.h>
#include <linux/swap.h>
#include <linux/swapops.h>
#include <linux/wait_bit.h>

static DEFINE_MUTEX(pgmap_lock);
static RADIX_TREE(pgmap_radix, GFP_KERNEL);
#define SECTION_MASK ~((1UL << PA_SECTION_SHIFT) - 1)
#define SECTION_SIZE (1UL << PA_SECTION_SHIFT)

static unsigned long order_at(struct resource *res, unsigned long pgoff)
{
	unsigned long phys_pgoff = PHYS_PFN(res->start) + pgoff;
	unsigned long nr_pages, mask;

	nr_pages = PHYS_PFN(resource_size(res));
	if (nr_pages == pgoff)
		return ULONG_MAX;

	/*
	 * What is the largest aligned power-of-2 range available from
	 * this resource pgoff to the end of the resource range,
	 * considering the alignment of the current pgoff?
	 */
	mask = phys_pgoff | rounddown_pow_of_two(nr_pages - pgoff);
	if (!mask)
		return ULONG_MAX;

	return find_first_bit(&mask, BITS_PER_LONG);
}

#define foreach_order_pgoff(res, order, pgoff) \
	for (pgoff = 0, order = order_at((res), pgoff); order < ULONG_MAX; \
			pgoff += 1UL << order, order = order_at((res), pgoff))

#if IS_ENABLED(CONFIG_DEVICE_PRIVATE)
vm_fault_t device_private_entry_fault(struct vm_area_struct *vma,
		       unsigned long addr,
		       swp_entry_t entry,
		       unsigned int flags,
		       pmd_t *pmdp)
{
	struct page *page = device_private_entry_to_page(entry);

	/*
	 * The page_fault() callback must migrate page back to system memory
	 * so that CPU can access it. This might fail for various reasons
	 * (device issue, device was unsafely unplugged, ...). When such
	 * error conditions happen, the callback must return VM_FAULT_SIGBUS.
	 *
	 * Note that because memory cgroup charges are accounted to the device
	 * memory, this should never fail because of memory restrictions (but
	 * allocation of regular system page might still fail because we are
	 * out of memory).
	 *
	 * There is a more in-depth description of what that callback can and
	 * cannot do, in include/linux/memremap.h
	 */
	return page->pgmap->page_fault(vma, addr, page, flags, pmdp);
}
EXPORT_SYMBOL(device_private_entry_fault);
#endif /* CONFIG_DEVICE_PRIVATE */

static void pgmap_radix_release(struct resource *res, unsigned long end_pgoff)
{
	unsigned long pgoff, order;

	mutex_lock(&pgmap_lock);
	foreach_order_pgoff(res, order, pgoff) {
		if (pgoff >= end_pgoff)
			break;
		radix_tree_delete(&pgmap_radix, PHYS_PFN(res->start) + pgoff);
	}
	mutex_unlock(&pgmap_lock);

	synchronize_rcu();
}

static unsigned long pfn_first(struct dev_pagemap *pgmap)
{
	const struct resource *res = &pgmap->res;
	struct vmem_altmap *altmap = &pgmap->altmap;
	unsigned long pfn;

	pfn = res->start >> PAGE_SHIFT;
	if (pgmap->altmap_valid)
		pfn += vmem_altmap_offset(altmap);
	return pfn;
}

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 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;

	for_each_device_pfn(pfn, pgmap)
		put_page(pfn_to_page(pfn));

	if (percpu_ref_tryget_live(pgmap->ref)) {
		dev_WARN(dev, "%s: page mapping is still live!\n", __func__);
		percpu_ref_put(pgmap->ref);
	}

	/* 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;

	mem_hotplug_begin();
	arch_remove_memory(align_start, align_size, pgmap->altmap_valid ?
			&pgmap->altmap : NULL);
	kasan_remove_zero_shadow(__va(align_start), align_size);
	mem_hotplug_done();

	untrack_pfn(NULL, PHYS_PFN(align_start), align_size);
	pgmap_radix_release(res, -1);
	dev_WARN_ONCE(dev, pgmap->altmap.alloc,
		      "%s: failed to free all reserved pages\n", __func__);
}

/**
 * devm_memremap_pages - remap and provide memmap backing for the given resource
 * @dev: hosting device for @res
 * @pgmap: pointer to a struct dev_pgmap
 *
 * Notes:
 * 1/ At a minimum the res, ref 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 altmap_valid
 *    must be set to true
 *
 * 3/ pgmap.ref must be 'live' on entry and 'dead' before devm_memunmap_pages()
 *    time (or devm release event). The expected order of events is that ref has
 *    been through percpu_ref_kill() before devm_memremap_pages_release(). The
 *    wait for the completion of all references being dropped and
 *    percpu_ref_exit() must occur after devm_memremap_pages_release().
 *
 * 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 vmem_altmap *altmap = pgmap->altmap_valid ?
			&pgmap->altmap : NULL;
	struct resource *res = &pgmap->res;
	unsigned long pfn, pgoff, order;
	pgprot_t pgprot = PAGE_KERNEL;
	int error, nid, is_ram;
	struct dev_pagemap *conflict_pgmap;

	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);
		return ERR_PTR(-ENOMEM);
	}

	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);
		return ERR_PTR(-ENOMEM);
	}

	is_ram = region_intersects(align_start, align_size,
		IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE);

	if (is_ram == REGION_MIXED) {
		WARN_ONCE(1, "%s attempted on mixed region %pr\n",
				__func__, res);
		return ERR_PTR(-ENXIO);
	}

	if (is_ram == REGION_INTERSECTS)
		return __va(res->start);

	if (!pgmap->ref)
		return ERR_PTR(-EINVAL);

	pgmap->dev = dev;

	mutex_lock(&pgmap_lock);
	error = 0;

	foreach_order_pgoff(res, order, pgoff) {
		error = __radix_tree_insert(&pgmap_radix,
				PHYS_PFN(res->start) + pgoff, order, pgmap);
		if (error) {
			dev_err(dev, "%s: failed: %d\n", __func__, error);
			break;
		}
	}
	mutex_unlock(&pgmap_lock);
	if (error)
		goto err_radix;

	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();
	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, altmap, false);
	if (!error)
		move_pfn_range_to_zone(&NODE_DATA(nid)->node_zones[ZONE_DEVICE],
					align_start >> PAGE_SHIFT,
					align_size >> PAGE_SHIFT, altmap);
	mem_hotplug_done();
	if (error)
		goto err_add_memory;

	for_each_device_pfn(pfn, pgmap) {
		struct page *page = pfn_to_page(pfn);

		/*
		 * ZONE_DEVICE pages union ->lru with a ->pgmap back
		 * pointer.  It is a bug if a ZONE_DEVICE page is ever
		 * freed or placed on a driver-private list.  Seed the
		 * storage with LIST_POISON* values.
		 */
		list_del(&page->lru);
		page->pgmap = pgmap;
		percpu_ref_get(pgmap->ref);
	}

	devm_add_action(dev, devm_memremap_pages_release, pgmap);

	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:
 err_radix:
	pgmap_radix_release(res, pgoff);
	return ERR_PTR(error);
}
EXPORT_SYMBOL(devm_memremap_pages);

unsigned long vmem_altmap_offset(struct vmem_altmap *altmap)
{
	/* number of pfns from base where pfn_to_page() is valid */
	return altmap->reserve + altmap->free;
}

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 = radix_tree_lookup(&pgmap_radix, 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
DEFINE_STATIC_KEY_FALSE(devmap_managed_key);
EXPORT_SYMBOL(devmap_managed_key);
static atomic_t devmap_enable;

/*
 * Toggle the static key for ->page_free() callbacks when dev_pagemap
 * pages go idle.
 */
void dev_pagemap_get_ops(void)
{
	if (atomic_inc_return(&devmap_enable) == 1)
		static_branch_enable(&devmap_managed_key);
}
EXPORT_SYMBOL_GPL(dev_pagemap_get_ops);

void dev_pagemap_put_ops(void)
{
	if (atomic_dec_and_test(&devmap_enable))
		static_branch_disable(&devmap_managed_key);
}
EXPORT_SYMBOL_GPL(dev_pagemap_put_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->page_free(page, page->pgmap->data);
	} 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. */
9476df7d	3	#include <linux/radix-tree.h>
7d3dcf26	4	#include <linux/device.h>
92281dee	5	#include <linux/types.h>
34c0fd54	6	#include <linux/pfn_t.h>
92281dee	7	#include <linux/io.h>
0207df4f	8	#include <linux/kasan.h>
92281dee	9	#include <linux/mm.h>
41e94a85	10	#include <linux/memory_hotplug.h>
5042db43 JG	11	#include <linux/swap.h>
5042db43 JG	12	#include <linux/swapops.h>
e7638488	13	#include <linux/wait_bit.h>
92281dee	14
9476df7d DW	15	static DEFINE_MUTEX(pgmap_lock);
	16	static RADIX_TREE(pgmap_radix, GFP_KERNEL);
	17	#define SECTION_MASK ~((1UL << PA_SECTION_SHIFT) - 1)
	18	#define SECTION_SIZE (1UL << PA_SECTION_SHIFT)
	19
ab1b597e	20	static unsigned long order_at(struct resource *res, unsigned long pgoff)
9476df7d	21	{
ab1b597e DW	22	unsigned long phys_pgoff = PHYS_PFN(res->start) + pgoff;
ab1b597e DW	23	unsigned long nr_pages, mask;
eb7d78c9	24
ab1b597e DW	25	nr_pages = PHYS_PFN(resource_size(res));
	26	if (nr_pages == pgoff)
	27	return ULONG_MAX;
	28
	29	/*
	30	* What is the largest aligned power-of-2 range available from
	31	* this resource pgoff to the end of the resource range,
	32	* considering the alignment of the current pgoff?
	33	*/
	34	mask = phys_pgoff \| rounddown_pow_of_two(nr_pages - pgoff);
	35	if (!mask)
	36	return ULONG_MAX;
	37
	38	return find_first_bit(&mask, BITS_PER_LONG);
	39	}
	40
	41	#define foreach_order_pgoff(res, order, pgoff) \
	42	for (pgoff = 0, order = order_at((res), pgoff); order < ULONG_MAX; \
	43	pgoff += 1UL << order, order = order_at((res), pgoff))
	44
5042db43	45	#if IS_ENABLED(CONFIG_DEVICE_PRIVATE)
2b740303	46	vm_fault_t device_private_entry_fault(struct vm_area_struct *vma,
5042db43 JG	47	unsigned long addr,
	48	swp_entry_t entry,
	49	unsigned int flags,
	50	pmd_t *pmdp)
	51	{
	52	struct page *page = device_private_entry_to_page(entry);
	53
	54	/*
	55	* The page_fault() callback must migrate page back to system memory
	56	* so that CPU can access it. This might fail for various reasons
	57	* (device issue, device was unsafely unplugged, ...). When such
	58	* error conditions happen, the callback must return VM_FAULT_SIGBUS.
	59	*
	60	* Note that because memory cgroup charges are accounted to the device
	61	* memory, this should never fail because of memory restrictions (but
	62	* allocation of regular system page might still fail because we are
	63	* out of memory).
	64	*
	65	* There is a more in-depth description of what that callback can and
	66	* cannot do, in include/linux/memremap.h
	67	*/
	68	return page->pgmap->page_fault(vma, addr, page, flags, pmdp);
	69	}
	70	EXPORT_SYMBOL(device_private_entry_fault);
	71	#endif /* CONFIG_DEVICE_PRIVATE */
	72
77dd66a3	73	static void pgmap_radix_release(struct resource *res, unsigned long end_pgoff)
ab1b597e DW	74	{
ab1b597e DW	75	unsigned long pgoff, order;
9476df7d DW	76
9476df7d DW	77	mutex_lock(&pgmap_lock);
77dd66a3 JS	78	foreach_order_pgoff(res, order, pgoff) {
	79	if (pgoff >= end_pgoff)
	80	break;
ab1b597e	81	radix_tree_delete(&pgmap_radix, PHYS_PFN(res->start) + pgoff);
77dd66a3	82	}
9476df7d	83	mutex_unlock(&pgmap_lock);
ab1b597e DW	84
ab1b597e DW	85	synchronize_rcu();
9476df7d DW	86	}
9476df7d DW	87
e7744aa2	88	static unsigned long pfn_first(struct dev_pagemap *pgmap)
5c2c2587	89	{
e7744aa2 LG	90	const struct resource *res = &pgmap->res;
e7744aa2 LG	91	struct vmem_altmap *altmap = &pgmap->altmap;
5c2c2587 DW	92	unsigned long pfn;
	93
	94	pfn = res->start >> PAGE_SHIFT;
e7744aa2	95	if (pgmap->altmap_valid)
5c2c2587 DW	96	pfn += vmem_altmap_offset(altmap);
	97	return pfn;
	98	}
	99
e7744aa2	100	static unsigned long pfn_end(struct dev_pagemap *pgmap)
5c2c2587	101	{
e7744aa2	102	const struct resource *res = &pgmap->res;
5c2c2587 DW	103
	104	return (res->start + resource_size(res)) >> PAGE_SHIFT;
	105	}
	106
949b9325 DW	107	static unsigned long pfn_next(unsigned long pfn)
	108	{
	109	if (pfn % 1024 == 0)
	110	cond_resched();
	111	return pfn + 1;
	112	}
	113
5c2c2587	114	#define for_each_device_pfn(pfn, map) \
949b9325	115	for (pfn = pfn_first(map); pfn < pfn_end(map); pfn = pfn_next(pfn))
5c2c2587	116
e8d51348	117	static void devm_memremap_pages_release(void *data)
41e94a85	118	{
e7744aa2	119	struct dev_pagemap *pgmap = data;
e8d51348	120	struct device *dev = pgmap->dev;
e7744aa2	121	struct resource *res = &pgmap->res;
9476df7d	122	resource_size_t align_start, align_size;
71389703 DW	123	unsigned long pfn;
71389703 DW	124
e7744aa2	125	for_each_device_pfn(pfn, pgmap)
71389703	126	put_page(pfn_to_page(pfn));
9476df7d	127
5c2c2587 DW	128	if (percpu_ref_tryget_live(pgmap->ref)) {
	129	dev_WARN(dev, "%s: page mapping is still live!\n", __func__);
	130	percpu_ref_put(pgmap->ref);
	131	}
	132
41e94a85	133	/* pages are dead and unused, undo the arch mapping */
9476df7d	134	align_start = res->start & ~(SECTION_SIZE - 1);
10a0cd6e JS	135	align_size = ALIGN(res->start + resource_size(res), SECTION_SIZE)
10a0cd6e JS	136	- align_start;
b5d24fda	137
f931ab47	138	mem_hotplug_begin();
e7744aa2 LG	139	arch_remove_memory(align_start, align_size, pgmap->altmap_valid ?
e7744aa2 LG	140	&pgmap->altmap : NULL);
0207df4f	141	kasan_remove_zero_shadow(__va(align_start), align_size);
f931ab47	142	mem_hotplug_done();
b5d24fda	143
9049771f	144	untrack_pfn(NULL, PHYS_PFN(align_start), align_size);
77dd66a3	145	pgmap_radix_release(res, -1);
e7744aa2 LG	146	dev_WARN_ONCE(dev, pgmap->altmap.alloc,
e7744aa2 LG	147	"%s: failed to free all reserved pages\n", __func__);
9476df7d DW	148	}
9476df7d DW	149
4b94ffdc DW	150	/**
	151	* devm_memremap_pages - remap and provide memmap backing for the given resource
	152	* @dev: hosting device for @res
e8d51348	153	* @pgmap: pointer to a struct dev_pgmap
4b94ffdc	154	*
5c2c2587	155	* Notes:
e8d51348 CH	156	* 1/ At a minimum the res, ref and type members of @pgmap must be initialized
	157	* by the caller before passing it to this function
	158	*
	159	* 2/ The altmap field may optionally be initialized, in which case altmap_valid
	160	* must be set to true
	161	*
	162	* 3/ pgmap.ref must be 'live' on entry and 'dead' before devm_memunmap_pages()
	163	* time (or devm release event). The expected order of events is that ref has
71389703 DW	164	* been through percpu_ref_kill() before devm_memremap_pages_release(). The
	165	* wait for the completion of all references being dropped and
	166	* percpu_ref_exit() must occur after devm_memremap_pages_release().
5c2c2587	167	*
e8d51348	168	* 4/ res is expected to be a host memory range that could feasibly be
5c2c2587 DW	169	* treated as a "System RAM" range, i.e. not a device mmio range, but
5c2c2587 DW	170	* this is not enforced.
4b94ffdc	171	*/
e8d51348	172	void devm_memremap_pages(struct device dev, struct dev_pagemap *pgmap)
41e94a85	173	{
ab1b597e	174	resource_size_t align_start, align_size, align_end;
e8d51348 CH	175	struct vmem_altmap *altmap = pgmap->altmap_valid ?
e8d51348 CH	176	&pgmap->altmap : NULL;
949b9325	177	struct resource *res = &pgmap->res;
ab1b597e	178	unsigned long pfn, pgoff, order;
9049771f	179	pgprot_t pgprot = PAGE_KERNEL;
949b9325	180	int error, nid, is_ram;
15d36fec	181	struct dev_pagemap *conflict_pgmap;
5f29a77c DW	182
	183	align_start = res->start & ~(SECTION_SIZE - 1);
	184	align_size = ALIGN(res->start + resource_size(res), SECTION_SIZE)
	185	- align_start;
15d36fec DJ	186	align_end = align_start + align_size - 1;
	187
	188	conflict_pgmap = get_dev_pagemap(PHYS_PFN(align_start), NULL);
	189	if (conflict_pgmap) {
	190	dev_WARN(dev, "Conflicting mapping in same section\n");
	191	put_dev_pagemap(conflict_pgmap);
	192	return ERR_PTR(-ENOMEM);
	193	}
	194
	195	conflict_pgmap = get_dev_pagemap(PHYS_PFN(align_end), NULL);
	196	if (conflict_pgmap) {
	197	dev_WARN(dev, "Conflicting mapping in same section\n");
	198	put_dev_pagemap(conflict_pgmap);
	199	return ERR_PTR(-ENOMEM);
	200	}
	201
d37a14bb LT	202	is_ram = region_intersects(align_start, align_size,
d37a14bb LT	203	IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE);
41e94a85 CH	204
	205	if (is_ram == REGION_MIXED) {
	206	WARN_ONCE(1, "%s attempted on mixed region %pr\n",
	207	__func__, res);
	208	return ERR_PTR(-ENXIO);
	209	}
	210
	211	if (is_ram == REGION_INTERSECTS)
	212	return __va(res->start);
	213
e8d51348	214	if (!pgmap->ref)
5c2c2587 DW	215	return ERR_PTR(-EINVAL);
5c2c2587 DW	216
4b94ffdc	217	pgmap->dev = dev;
4b94ffdc	218
9476df7d DW	219	mutex_lock(&pgmap_lock);
9476df7d DW	220	error = 0;
ab1b597e DW	221
ab1b597e DW	222	foreach_order_pgoff(res, order, pgoff) {
ab1b597e	223	error = __radix_tree_insert(&pgmap_radix,
e7744aa2	224	PHYS_PFN(res->start) + pgoff, order, pgmap);
9476df7d DW	225	if (error) {
	226	dev_err(dev, "%s: failed: %d\n", __func__, error);
	227	break;
	228	}
	229	}
	230	mutex_unlock(&pgmap_lock);
	231	if (error)
	232	goto err_radix;
	233
41e94a85 CH	234	nid = dev_to_node(dev);
41e94a85 CH	235	if (nid < 0)
7eff93b7	236	nid = numa_mem_id();
41e94a85	237
9049771f DW	238	error = track_pfn_remap(NULL, &pgprot, PHYS_PFN(align_start), 0,
	239	align_size);
	240	if (error)
	241	goto err_pfn_remap;
	242
f931ab47	243	mem_hotplug_begin();
0207df4f AR	244	error = kasan_add_zero_shadow(__va(align_start), align_size);
	245	if (error) {
	246	mem_hotplug_done();
	247	goto err_kasan;
	248	}
	249
24e6d5a5	250	error = arch_add_memory(nid, align_start, align_size, altmap, false);
f1dd2cd1 MH	251	if (!error)
	252	move_pfn_range_to_zone(&NODE_DATA(nid)->node_zones[ZONE_DEVICE],
	253	align_start >> PAGE_SHIFT,
a99583e7	254	align_size >> PAGE_SHIFT, altmap);
f931ab47	255	mem_hotplug_done();
9476df7d DW	256	if (error)
9476df7d DW	257	goto err_add_memory;
41e94a85	258
e7744aa2	259	for_each_device_pfn(pfn, pgmap) {
5c2c2587 DW	260	struct page *page = pfn_to_page(pfn);
5c2c2587 DW	261
d77a117e DW	262	/*
	263	* ZONE_DEVICE pages union ->lru with a ->pgmap back
	264	* pointer. It is a bug if a ZONE_DEVICE page is ever
	265	* freed or placed on a driver-private list. Seed the
	266	* storage with LIST_POISON* values.
	267	*/
	268	list_del(&page->lru);
5c2c2587	269	page->pgmap = pgmap;
e8d51348	270	percpu_ref_get(pgmap->ref);
5c2c2587	271	}
e8d51348 CH	272
	273	devm_add_action(dev, devm_memremap_pages_release, pgmap);
	274
41e94a85	275	return __va(res->start);
9476df7d DW	276
9476df7d DW	277	err_add_memory:
0207df4f AR	278	kasan_remove_zero_shadow(__va(align_start), align_size);
0207df4f AR	279	err_kasan:
9049771f DW	280	untrack_pfn(NULL, PHYS_PFN(align_start), align_size);
9049771f DW	281	err_pfn_remap:
9476df7d	282	err_radix:
77dd66a3	283	pgmap_radix_release(res, pgoff);
9476df7d	284	return ERR_PTR(error);
41e94a85 CH	285	}
41e94a85 CH	286	EXPORT_SYMBOL(devm_memremap_pages);
4b94ffdc DW	287
	288	unsigned long vmem_altmap_offset(struct vmem_altmap *altmap)
	289	{
	290	/* number of pfns from base where pfn_to_page() is valid */
	291	return altmap->reserve + altmap->free;
	292	}
	293
	294	void vmem_altmap_free(struct vmem_altmap *altmap, unsigned long nr_pfns)
	295	{
	296	altmap->alloc -= nr_pfns;
	297	}
	298
0822acb8 CH	299	/**
	300	* get_dev_pagemap() - take a new live reference on the dev_pagemap for @pfn
	301	* @pfn: page frame number to lookup page_map
	302	* @pgmap: optional known pgmap that already has a reference
	303	*
832d7aa0 CH	304	* If @pgmap is non-NULL and covers @pfn it will be returned as-is. If @pgmap
832d7aa0 CH	305	* is non-NULL but does not cover @pfn the reference to it will be released.
0822acb8 CH	306	*/
	307	struct dev_pagemap *get_dev_pagemap(unsigned long pfn,
	308	struct dev_pagemap *pgmap)
	309	{
0822acb8 CH	310	resource_size_t phys = PFN_PHYS(pfn);
	311
	312	/*
832d7aa0	313	* In the cached case we're already holding a live reference.
0822acb8	314	*/
832d7aa0	315	if (pgmap) {
e7744aa2	316	if (phys >= pgmap->res.start && phys <= pgmap->res.end)
832d7aa0 CH	317	return pgmap;
832d7aa0 CH	318	put_dev_pagemap(pgmap);
0822acb8 CH	319	}
	320
	321	/* fall back to slow path lookup */
	322	rcu_read_lock();
e697c5b9	323	pgmap = radix_tree_lookup(&pgmap_radix, PHYS_PFN(phys));
0822acb8 CH	324	if (pgmap && !percpu_ref_tryget_live(pgmap->ref))
	325	pgmap = NULL;
	326	rcu_read_unlock();
	327
	328	return pgmap;
	329	}
e7638488	330	EXPORT_SYMBOL_GPL(get_dev_pagemap);
7b2d55d2	331
e7638488 DW	332	#ifdef CONFIG_DEV_PAGEMAP_OPS
e7638488 DW	333	DEFINE_STATIC_KEY_FALSE(devmap_managed_key);
31c5bda3	334	EXPORT_SYMBOL(devmap_managed_key);
e7638488 DW	335	static atomic_t devmap_enable;
	336
	337	/*
	338	* Toggle the static key for ->page_free() callbacks when dev_pagemap
	339	* pages go idle.
	340	*/
	341	void dev_pagemap_get_ops(void)
	342	{
	343	if (atomic_inc_return(&devmap_enable) == 1)
	344	static_branch_enable(&devmap_managed_key);
	345	}
	346	EXPORT_SYMBOL_GPL(dev_pagemap_get_ops);
	347
	348	void dev_pagemap_put_ops(void)
	349	{
	350	if (atomic_dec_and_test(&devmap_enable))
	351	static_branch_disable(&devmap_managed_key);
	352	}
	353	EXPORT_SYMBOL_GPL(dev_pagemap_put_ops);
	354
	355	void __put_devmap_managed_page(struct page *page)
7b2d55d2 JG	356	{
	357	int count = page_ref_dec_return(page);
	358
	359	/*
	360	* If refcount is 1 then page is freed and refcount is stable as nobody
	361	* holds a reference on the page.
	362	*/
	363	if (count == 1) {
	364	/* Clear Active bit in case of parallel mark_page_accessed */
	365	__ClearPageActive(page);
	366	__ClearPageWaiters(page);
	367
c733a828	368	mem_cgroup_uncharge(page);
7b2d55d2 JG	369
	370	page->pgmap->page_free(page, page->pgmap->data);
	371	} else if (!count)
	372	__put_page(page);
	373	}
31c5bda3	374	EXPORT_SYMBOL(__put_devmap_managed_page);
e7638488	375	#endif /* CONFIG_DEV_PAGEMAP_OPS */