[linux.git] / kernel / memremap.c

/*
 * Copyright(c) 2015 Intel Corporation. All rights reserved.
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of version 2 of the GNU General Public License as
 * published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful, but
 * WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 */
#include <linux/radix-tree.h>
#include <linux/memremap.h>
#include <linux/device.h>
#include <linux/types.h>
#include <linux/pfn_t.h>
#include <linux/io.h>
#include <linux/mm.h>
#include <linux/memory_hotplug.h>

#ifndef ioremap_cache
/* temporary while we convert existing ioremap_cache users to memremap */
__weak void __iomem *ioremap_cache(resource_size_t offset, unsigned long size)
{
	return ioremap(offset, size);
}
#endif

#ifndef arch_memremap_wb
static void *arch_memremap_wb(resource_size_t offset, unsigned long size)
{
	return (__force void *)ioremap_cache(offset, size);
}
#endif

static void *try_ram_remap(resource_size_t offset, size_t size)
{
	unsigned long pfn = PHYS_PFN(offset);

	/* In the simple case just return the existing linear address */
	if (pfn_valid(pfn) && !PageHighMem(pfn_to_page(pfn)))
		return __va(offset);
	return NULL; /* fallback to arch_memremap_wb */
}

/**
 * memremap() - remap an iomem_resource as cacheable memory
 * @offset: iomem resource start address
 * @size: size of remap
 * @flags: any of MEMREMAP_WB, MEMREMAP_WT and MEMREMAP_WC
 *
 * memremap() is "ioremap" for cases where it is known that the resource
 * being mapped does not have i/o side effects and the __iomem
 * annotation is not applicable. In the case of multiple flags, the different
 * mapping types will be attempted in the order listed below until one of
 * them succeeds.
 *
 * MEMREMAP_WB - matches the default mapping for System RAM on
 * the architecture.  This is usually a read-allocate write-back cache.
 * Morever, if MEMREMAP_WB is specified and the requested remap region is RAM
 * memremap() will bypass establishing a new mapping and instead return
 * a pointer into the direct map.
 *
 * MEMREMAP_WT - establish a mapping whereby writes either bypass the
 * cache or are written through to memory and never exist in a
 * cache-dirty state with respect to program visibility.  Attempts to
 * map System RAM with this mapping type will fail.
 *
 * MEMREMAP_WC - establish a writecombine mapping, whereby writes may
 * be coalesced together (e.g. in the CPU's write buffers), but is otherwise
 * uncached. Attempts to map System RAM with this mapping type will fail.
 */
void *memremap(resource_size_t offset, size_t size, unsigned long flags)
{
	int is_ram = region_intersects(offset, size,
				       IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE);
	void *addr = NULL;

	if (!flags)
		return NULL;

	if (is_ram == REGION_MIXED) {
		WARN_ONCE(1, "memremap attempted on mixed range %pa size: %#lx\n",
				&offset, (unsigned long) size);
		return NULL;
	}

	/* Try all mapping types requested until one returns non-NULL */
	if (flags & MEMREMAP_WB) {
		/*
		 * MEMREMAP_WB is special in that it can be satisifed
		 * from the direct map.  Some archs depend on the
		 * capability of memremap() to autodetect cases where
		 * the requested range is potentially in System RAM.
		 */
		if (is_ram == REGION_INTERSECTS)
			addr = try_ram_remap(offset, size);
		if (!addr)
			addr = arch_memremap_wb(offset, size);
	}

	/*
	 * If we don't have a mapping yet and other request flags are
	 * present then we will be attempting to establish a new virtual
	 * address mapping.  Enforce that this mapping is not aliasing
	 * System RAM.
	 */
	if (!addr && is_ram == REGION_INTERSECTS && flags != MEMREMAP_WB) {
		WARN_ONCE(1, "memremap attempted on ram %pa size: %#lx\n",
				&offset, (unsigned long) size);
		return NULL;
	}

	if (!addr && (flags & MEMREMAP_WT))
		addr = ioremap_wt(offset, size);

	if (!addr && (flags & MEMREMAP_WC))
		addr = ioremap_wc(offset, size);

	return addr;
}
EXPORT_SYMBOL(memremap);

void memunmap(void *addr)
{
	if (is_vmalloc_addr(addr))
		iounmap((void __iomem *) addr);
}
EXPORT_SYMBOL(memunmap);

static void devm_memremap_release(struct device *dev, void *res)
{
	memunmap(*(void **)res);
}

static int devm_memremap_match(struct device *dev, void *res, void *match_data)
{
	return *(void **)res == match_data;
}

void *devm_memremap(struct device *dev, resource_size_t offset,
		size_t size, unsigned long flags)
{
	void **ptr, *addr;

	ptr = devres_alloc_node(devm_memremap_release, sizeof(*ptr), GFP_KERNEL,
			dev_to_node(dev));
	if (!ptr)
		return ERR_PTR(-ENOMEM);

	addr = memremap(offset, size, flags);
	if (addr) {
		*ptr = addr;
		devres_add(dev, ptr);
	} else {
		devres_free(ptr);
		return ERR_PTR(-ENXIO);
	}

	return addr;
}
EXPORT_SYMBOL(devm_memremap);

void devm_memunmap(struct device *dev, void *addr)
{
	WARN_ON(devres_release(dev, devm_memremap_release,
				devm_memremap_match, addr));
}
EXPORT_SYMBOL(devm_memunmap);

#ifdef CONFIG_ZONE_DEVICE
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)

struct page_map {
	struct resource res;
	struct percpu_ref *ref;
	struct dev_pagemap pgmap;
	struct vmem_altmap altmap;
};

void get_zone_device_page(struct page *page)
{
	percpu_ref_get(page->pgmap->ref);
}
EXPORT_SYMBOL(get_zone_device_page);

void put_zone_device_page(struct page *page)
{
	put_dev_pagemap(page->pgmap);
}
EXPORT_SYMBOL(put_zone_device_page);

static void pgmap_radix_release(struct resource *res)
{
	resource_size_t key, align_start, align_size, align_end;

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

	mutex_lock(&pgmap_lock);
	for (key = res->start; key <= res->end; key += SECTION_SIZE)
		radix_tree_delete(&pgmap_radix, key >> PA_SECTION_SHIFT);
	mutex_unlock(&pgmap_lock);
}

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

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

static unsigned long pfn_end(struct page_map *page_map)
{
	const struct resource *res = &page_map->res;

	return (res->start + resource_size(res)) >> PAGE_SHIFT;
}

#define for_each_device_pfn(pfn, map) \
	for (pfn = pfn_first(map); pfn < pfn_end(map); pfn++)

static void devm_memremap_pages_release(struct device *dev, void *data)
{
	struct page_map *page_map = data;
	struct resource *res = &page_map->res;
	resource_size_t align_start, align_size;
	struct dev_pagemap *pgmap = &page_map->pgmap;

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

/* assumes rcu_read_lock() held at entry */
struct dev_pagemap *find_dev_pagemap(resource_size_t phys)
{
	struct page_map *page_map;

	WARN_ON_ONCE(!rcu_read_lock_held());

	page_map = radix_tree_lookup(&pgmap_radix, phys >> PA_SECTION_SHIFT);
	return page_map ? &page_map->pgmap : NULL;
}

/**
 * devm_memremap_pages - remap and provide memmap backing for the given resource
 * @dev: hosting device for @res
 * @res: "host memory" address range
 * @ref: a live per-cpu reference count
 * @altmap: optional descriptor for allocating the memmap from @res
 *
 * Notes:
 * 1/ @ref must be 'live' on entry and 'dead' before devm_memunmap_pages() time
 *    (or devm release event).
 *
 * 2/ @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 resource *res,
		struct percpu_ref *ref, struct vmem_altmap *altmap)
{
	resource_size_t key, align_start, align_size, align_end;
	pgprot_t pgprot = PAGE_KERNEL;
	struct dev_pagemap *pgmap;
	struct page_map *page_map;
	int error, nid, is_ram;
	unsigned long pfn;

	align_start = res->start & ~(SECTION_SIZE - 1);
	align_size = ALIGN(res->start + resource_size(res), SECTION_SIZE)
		- align_start;
	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 (!ref)
		return ERR_PTR(-EINVAL);

	page_map = devres_alloc_node(devm_memremap_pages_release,
			sizeof(*page_map), GFP_KERNEL, dev_to_node(dev));
	if (!page_map)
		return ERR_PTR(-ENOMEM);
	pgmap = &page_map->pgmap;

	memcpy(&page_map->res, res, sizeof(*res));

	pgmap->dev = dev;
	if (altmap) {
		memcpy(&page_map->altmap, altmap, sizeof(*altmap));
		pgmap->altmap = &page_map->altmap;
	}
	pgmap->ref = ref;
	pgmap->res = &page_map->res;

	mutex_lock(&pgmap_lock);
	error = 0;
	align_end = align_start + align_size - 1;
	for (key = align_start; key <= align_end; key += SECTION_SIZE) {
		struct dev_pagemap *dup;

		rcu_read_lock();
		dup = find_dev_pagemap(key);
		rcu_read_unlock();
		if (dup) {
			dev_err(dev, "%s: %pr collides with mapping for %s\n",
					__func__, res, dev_name(dup->dev));
			error = -EBUSY;
			break;
		}
		error = radix_tree_insert(&pgmap_radix, key >> PA_SECTION_SHIFT,
				page_map);
		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;

	error = arch_add_memory(nid, align_start, align_size, true);
	if (error)
		goto err_add_memory;

	for_each_device_pfn(pfn, page_map) {
		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;
	}
	devres_add(dev, page_map);
	return __va(res->start);

 err_add_memory:
	untrack_pfn(NULL, PHYS_PFN(align_start), align_size);
 err_pfn_remap:
 err_radix:
	pgmap_radix_release(res);
	devres_free(page_map);
	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;
}

struct vmem_altmap *to_vmem_altmap(unsigned long memmap_start)
{
	/*
	 * 'memmap_start' is the virtual address for the first "struct
	 * page" in this range of the vmemmap array.  In the case of
	 * CONFIG_SPARSEMEM_VMEMMAP a page_to_pfn conversion is simple
	 * pointer arithmetic, so we can perform this to_vmem_altmap()
	 * conversion without concern for the initialization state of
	 * the struct page fields.
	 */
	struct page *page = (struct page *) memmap_start;
	struct dev_pagemap *pgmap;

	/*
	 * Unconditionally retrieve a dev_pagemap associated with the
	 * given physical address, this is only for use in the
	 * arch_{add|remove}_memory() for setting up and tearing down
	 * the memmap.
	 */
	rcu_read_lock();
	pgmap = find_dev_pagemap(__pfn_to_phys(page_to_pfn(page)));
	rcu_read_unlock();

	return pgmap ? pgmap->altmap : NULL;
}
#endif /* CONFIG_ZONE_DEVICE */
Commit	Line	Data
92281dee DW	1	/*
	2	* Copyright(c) 2015 Intel Corporation. All rights reserved.
	3	*
	4	* This program is free software; you can redistribute it and/or modify
	5	* it under the terms of version 2 of the GNU General Public License as
	6	* published by the Free Software Foundation.
	7	*
	8	* This program is distributed in the hope that it will be useful, but
	9	* WITHOUT ANY WARRANTY; without even the implied warranty of
	10	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
	11	* General Public License for more details.
	12	*/
9476df7d DW	13	#include <linux/radix-tree.h>
9476df7d DW	14	#include <linux/memremap.h>
7d3dcf26	15	#include <linux/device.h>
92281dee	16	#include <linux/types.h>
34c0fd54	17	#include <linux/pfn_t.h>
92281dee DW	18	#include <linux/io.h>
92281dee DW	19	#include <linux/mm.h>
41e94a85	20	#include <linux/memory_hotplug.h>
92281dee DW	21
	22	#ifndef ioremap_cache
	23	/* temporary while we convert existing ioremap_cache users to memremap */
	24	__weak void __iomem *ioremap_cache(resource_size_t offset, unsigned long size)
	25	{
	26	return ioremap(offset, size);
	27	}
	28	#endif
	29
c269cba3 AB	30	#ifndef arch_memremap_wb
	31	static void *arch_memremap_wb(resource_size_t offset, unsigned long size)
	32	{
	33	return (__force void *)ioremap_cache(offset, size);
	34	}
	35	#endif
	36
182475b7 DW	37	static void *try_ram_remap(resource_size_t offset, size_t size)
182475b7 DW	38	{
ac343e88	39	unsigned long pfn = PHYS_PFN(offset);
182475b7 DW	40
182475b7 DW	41	/* In the simple case just return the existing linear address */
ac343e88	42	if (pfn_valid(pfn) && !PageHighMem(pfn_to_page(pfn)))
182475b7	43	return __va(offset);
c269cba3	44	return NULL; /* fallback to arch_memremap_wb */
182475b7 DW	45	}
182475b7 DW	46
92281dee DW	47	/**
	48	* memremap() - remap an iomem_resource as cacheable memory
	49	* @offset: iomem resource start address
	50	* @size: size of remap
c907e0eb	51	* @flags: any of MEMREMAP_WB, MEMREMAP_WT and MEMREMAP_WC
92281dee DW	52	*
	53	* memremap() is "ioremap" for cases where it is known that the resource
	54	* being mapped does not have i/o side effects and the __iomem
c907e0eb BS	55	* annotation is not applicable. In the case of multiple flags, the different
	56	* mapping types will be attempted in the order listed below until one of
	57	* them succeeds.
92281dee	58	*
1c29f25b	59	* MEMREMAP_WB - matches the default mapping for System RAM on
92281dee DW	60	* the architecture. This is usually a read-allocate write-back cache.
	61	* Morever, if MEMREMAP_WB is specified and the requested remap region is RAM
	62	* memremap() will bypass establishing a new mapping and instead return
	63	* a pointer into the direct map.
	64	*
	65	* MEMREMAP_WT - establish a mapping whereby writes either bypass the
	66	* cache or are written through to memory and never exist in a
	67	* cache-dirty state with respect to program visibility. Attempts to
1c29f25b	68	* map System RAM with this mapping type will fail.
c907e0eb BS	69	*
	70	* MEMREMAP_WC - establish a writecombine mapping, whereby writes may
	71	* be coalesced together (e.g. in the CPU's write buffers), but is otherwise
	72	* uncached. Attempts to map System RAM with this mapping type will fail.
92281dee DW	73	*/
	74	void *memremap(resource_size_t offset, size_t size, unsigned long flags)
	75	{
1c29f25b TK	76	int is_ram = region_intersects(offset, size,
1c29f25b TK	77	IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE);
92281dee DW	78	void *addr = NULL;
92281dee DW	79
cf61e2a1 BS	80	if (!flags)
	81	return NULL;
	82
92281dee DW	83	if (is_ram == REGION_MIXED) {
	84	WARN_ONCE(1, "memremap attempted on mixed range %pa size: %#lx\n",
	85	&offset, (unsigned long) size);
	86	return NULL;
	87	}
	88
	89	/* Try all mapping types requested until one returns non-NULL */
	90	if (flags & MEMREMAP_WB) {
92281dee DW	91	/*
	92	* MEMREMAP_WB is special in that it can be satisifed
	93	* from the direct map. Some archs depend on the
	94	* capability of memremap() to autodetect cases where
1c29f25b	95	* the requested range is potentially in System RAM.
92281dee DW	96	*/
92281dee DW	97	if (is_ram == REGION_INTERSECTS)
182475b7 DW	98	addr = try_ram_remap(offset, size);
182475b7 DW	99	if (!addr)
c269cba3	100	addr = arch_memremap_wb(offset, size);
92281dee DW	101	}
	102
	103	/*
cf61e2a1 BS	104	* If we don't have a mapping yet and other request flags are
cf61e2a1 BS	105	* present then we will be attempting to establish a new virtual
92281dee	106	* address mapping. Enforce that this mapping is not aliasing
1c29f25b	107	* System RAM.
92281dee	108	*/
cf61e2a1	109	if (!addr && is_ram == REGION_INTERSECTS && flags != MEMREMAP_WB) {
92281dee DW	110	WARN_ONCE(1, "memremap attempted on ram %pa size: %#lx\n",
	111	&offset, (unsigned long) size);
	112	return NULL;
	113	}
	114
cf61e2a1	115	if (!addr && (flags & MEMREMAP_WT))
92281dee	116	addr = ioremap_wt(offset, size);
c907e0eb BS	117
	118	if (!addr && (flags & MEMREMAP_WC))
	119	addr = ioremap_wc(offset, size);
92281dee DW	120
	121	return addr;
	122	}
	123	EXPORT_SYMBOL(memremap);
	124
	125	void memunmap(void *addr)
	126	{
	127	if (is_vmalloc_addr(addr))
	128	iounmap((void __iomem *) addr);
	129	}
	130	EXPORT_SYMBOL(memunmap);
7d3dcf26 CH	131
	132	static void devm_memremap_release(struct device dev, void res)
	133	{
9273a8bb	134	memunmap((void *)res);
7d3dcf26 CH	135	}
	136
	137	static int devm_memremap_match(struct device dev, void res, void *match_data)
	138	{
	139	return (void *)res == match_data;
	140	}
	141
	142	void devm_memremap(struct device dev, resource_size_t offset,
	143	size_t size, unsigned long flags)
	144	{
	145	void *ptr, addr;
	146
538ea4aa DW	147	ptr = devres_alloc_node(devm_memremap_release, sizeof(*ptr), GFP_KERNEL,
538ea4aa DW	148	dev_to_node(dev));
7d3dcf26	149	if (!ptr)
b36f4761	150	return ERR_PTR(-ENOMEM);
7d3dcf26 CH	151
	152	addr = memremap(offset, size, flags);
	153	if (addr) {
	154	*ptr = addr;
	155	devres_add(dev, ptr);
93f834df	156	} else {
7d3dcf26	157	devres_free(ptr);
93f834df TK	158	return ERR_PTR(-ENXIO);
93f834df TK	159	}
7d3dcf26 CH	160
	161	return addr;
	162	}
	163	EXPORT_SYMBOL(devm_memremap);
	164
	165	void devm_memunmap(struct device dev, void addr)
	166	{
d741314f DW	167	WARN_ON(devres_release(dev, devm_memremap_release,
d741314f DW	168	devm_memremap_match, addr));
7d3dcf26 CH	169	}
7d3dcf26 CH	170	EXPORT_SYMBOL(devm_memunmap);
41e94a85 CH	171
41e94a85 CH	172	#ifdef CONFIG_ZONE_DEVICE
9476df7d DW	173	static DEFINE_MUTEX(pgmap_lock);
	174	static RADIX_TREE(pgmap_radix, GFP_KERNEL);
	175	#define SECTION_MASK ~((1UL << PA_SECTION_SHIFT) - 1)
	176	#define SECTION_SIZE (1UL << PA_SECTION_SHIFT)
	177
41e94a85 CH	178	struct page_map {
41e94a85 CH	179	struct resource res;
9476df7d DW	180	struct percpu_ref *ref;
9476df7d DW	181	struct dev_pagemap pgmap;
4b94ffdc	182	struct vmem_altmap altmap;
41e94a85 CH	183	};
41e94a85 CH	184
3565fce3 DW	185	void get_zone_device_page(struct page *page)
	186	{
	187	percpu_ref_get(page->pgmap->ref);
	188	}
	189	EXPORT_SYMBOL(get_zone_device_page);
	190
	191	void put_zone_device_page(struct page *page)
	192	{
	193	put_dev_pagemap(page->pgmap);
	194	}
	195	EXPORT_SYMBOL(put_zone_device_page);
	196
9476df7d DW	197	static void pgmap_radix_release(struct resource *res)
9476df7d DW	198	{
eb7d78c9 DW	199	resource_size_t key, align_start, align_size, align_end;
	200
	201	align_start = res->start & ~(SECTION_SIZE - 1);
	202	align_size = ALIGN(resource_size(res), SECTION_SIZE);
	203	align_end = align_start + align_size - 1;
9476df7d DW	204
	205	mutex_lock(&pgmap_lock);
	206	for (key = res->start; key <= res->end; key += SECTION_SIZE)
	207	radix_tree_delete(&pgmap_radix, key >> PA_SECTION_SHIFT);
	208	mutex_unlock(&pgmap_lock);
	209	}
	210
5c2c2587 DW	211	static unsigned long pfn_first(struct page_map *page_map)
	212	{
	213	struct dev_pagemap *pgmap = &page_map->pgmap;
	214	const struct resource *res = &page_map->res;
	215	struct vmem_altmap *altmap = pgmap->altmap;
	216	unsigned long pfn;
	217
	218	pfn = res->start >> PAGE_SHIFT;
	219	if (altmap)
	220	pfn += vmem_altmap_offset(altmap);
	221	return pfn;
	222	}
	223
	224	static unsigned long pfn_end(struct page_map *page_map)
	225	{
	226	const struct resource *res = &page_map->res;
	227
	228	return (res->start + resource_size(res)) >> PAGE_SHIFT;
	229	}
	230
	231	#define for_each_device_pfn(pfn, map) \
	232	for (pfn = pfn_first(map); pfn < pfn_end(map); pfn++)
	233
9476df7d	234	static void devm_memremap_pages_release(struct device dev, void data)
41e94a85	235	{
9476df7d DW	236	struct page_map *page_map = data;
	237	struct resource *res = &page_map->res;
	238	resource_size_t align_start, align_size;
4b94ffdc	239	struct dev_pagemap *pgmap = &page_map->pgmap;
9476df7d	240
5c2c2587 DW	241	if (percpu_ref_tryget_live(pgmap->ref)) {
	242	dev_WARN(dev, "%s: page mapping is still live!\n", __func__);
	243	percpu_ref_put(pgmap->ref);
	244	}
	245
41e94a85	246	/* pages are dead and unused, undo the arch mapping */
9476df7d DW	247	align_start = res->start & ~(SECTION_SIZE - 1);
	248	align_size = ALIGN(resource_size(res), SECTION_SIZE);
	249	arch_remove_memory(align_start, align_size);
9049771f	250	untrack_pfn(NULL, PHYS_PFN(align_start), align_size);
eb7d78c9	251	pgmap_radix_release(res);
4b94ffdc DW	252	dev_WARN_ONCE(dev, pgmap->altmap && pgmap->altmap->alloc,
4b94ffdc DW	253	"%s: failed to free all reserved pages\n", __func__);
9476df7d DW	254	}
	255
	256	/* assumes rcu_read_lock() held at entry */
	257	struct dev_pagemap *find_dev_pagemap(resource_size_t phys)
	258	{
	259	struct page_map *page_map;
	260
	261	WARN_ON_ONCE(!rcu_read_lock_held());
	262
	263	page_map = radix_tree_lookup(&pgmap_radix, phys >> PA_SECTION_SHIFT);
	264	return page_map ? &page_map->pgmap : NULL;
41e94a85 CH	265	}
41e94a85 CH	266
4b94ffdc DW	267	/**
	268	* devm_memremap_pages - remap and provide memmap backing for the given resource
	269	* @dev: hosting device for @res
	270	* @res: "host memory" address range
5c2c2587	271	* @ref: a live per-cpu reference count
4b94ffdc DW	272	* @altmap: optional descriptor for allocating the memmap from @res
4b94ffdc DW	273	*
5c2c2587 DW	274	* Notes:
	275	* 1/ @ref must be 'live' on entry and 'dead' before devm_memunmap_pages() time
	276	* (or devm release event).
	277	*
	278	* 2/ @res is expected to be a host memory range that could feasibly be
	279	* treated as a "System RAM" range, i.e. not a device mmio range, but
	280	* this is not enforced.
4b94ffdc DW	281	*/
4b94ffdc DW	282	void devm_memremap_pages(struct device dev, struct resource *res,
5c2c2587	283	struct percpu_ref ref, struct vmem_altmap altmap)
41e94a85	284	{
eb7d78c9	285	resource_size_t key, align_start, align_size, align_end;
9049771f	286	pgprot_t pgprot = PAGE_KERNEL;
4b94ffdc	287	struct dev_pagemap *pgmap;
41e94a85	288	struct page_map *page_map;
5f29a77c	289	int error, nid, is_ram;
5c2c2587	290	unsigned long pfn;
5f29a77c DW	291
	292	align_start = res->start & ~(SECTION_SIZE - 1);
	293	align_size = ALIGN(res->start + resource_size(res), SECTION_SIZE)
	294	- align_start;
d37a14bb LT	295	is_ram = region_intersects(align_start, align_size,
d37a14bb LT	296	IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE);
41e94a85 CH	297
	298	if (is_ram == REGION_MIXED) {
	299	WARN_ONCE(1, "%s attempted on mixed region %pr\n",
	300	__func__, res);
	301	return ERR_PTR(-ENXIO);
	302	}
	303
	304	if (is_ram == REGION_INTERSECTS)
	305	return __va(res->start);
	306
5c2c2587 DW	307	if (!ref)
	308	return ERR_PTR(-EINVAL);
	309
538ea4aa DW	310	page_map = devres_alloc_node(devm_memremap_pages_release,
538ea4aa DW	311	sizeof(*page_map), GFP_KERNEL, dev_to_node(dev));
41e94a85 CH	312	if (!page_map)
41e94a85 CH	313	return ERR_PTR(-ENOMEM);
4b94ffdc	314	pgmap = &page_map->pgmap;
41e94a85 CH	315
	316	memcpy(&page_map->res, res, sizeof(*res));
	317
4b94ffdc DW	318	pgmap->dev = dev;
	319	if (altmap) {
	320	memcpy(&page_map->altmap, altmap, sizeof(*altmap));
	321	pgmap->altmap = &page_map->altmap;
	322	}
5c2c2587	323	pgmap->ref = ref;
4b94ffdc DW	324	pgmap->res = &page_map->res;
4b94ffdc DW	325
9476df7d DW	326	mutex_lock(&pgmap_lock);
9476df7d DW	327	error = 0;
eb7d78c9 DW	328	align_end = align_start + align_size - 1;
eb7d78c9 DW	329	for (key = align_start; key <= align_end; key += SECTION_SIZE) {
9476df7d DW	330	struct dev_pagemap *dup;
	331
	332	rcu_read_lock();
	333	dup = find_dev_pagemap(key);
	334	rcu_read_unlock();
	335	if (dup) {
	336	dev_err(dev, "%s: %pr collides with mapping for %s\n",
	337	__func__, res, dev_name(dup->dev));
	338	error = -EBUSY;
	339	break;
	340	}
	341	error = radix_tree_insert(&pgmap_radix, key >> PA_SECTION_SHIFT,
	342	page_map);
	343	if (error) {
	344	dev_err(dev, "%s: failed: %d\n", __func__, error);
	345	break;
	346	}
	347	}
	348	mutex_unlock(&pgmap_lock);
	349	if (error)
	350	goto err_radix;
	351
41e94a85 CH	352	nid = dev_to_node(dev);
41e94a85 CH	353	if (nid < 0)
7eff93b7	354	nid = numa_mem_id();
41e94a85	355
9049771f DW	356	error = track_pfn_remap(NULL, &pgprot, PHYS_PFN(align_start), 0,
	357	align_size);
	358	if (error)
	359	goto err_pfn_remap;
	360
9476df7d DW	361	error = arch_add_memory(nid, align_start, align_size, true);
	362	if (error)
	363	goto err_add_memory;
41e94a85	364
5c2c2587 DW	365	for_each_device_pfn(pfn, page_map) {
	366	struct page *page = pfn_to_page(pfn);
	367
d77a117e DW	368	/*
	369	* ZONE_DEVICE pages union ->lru with a ->pgmap back
	370	* pointer. It is a bug if a ZONE_DEVICE page is ever
	371	* freed or placed on a driver-private list. Seed the
	372	* storage with LIST_POISON* values.
	373	*/
	374	list_del(&page->lru);
5c2c2587 DW	375	page->pgmap = pgmap;
5c2c2587 DW	376	}
41e94a85 CH	377	devres_add(dev, page_map);
41e94a85 CH	378	return __va(res->start);
9476df7d DW	379
9476df7d DW	380	err_add_memory:
9049771f DW	381	untrack_pfn(NULL, PHYS_PFN(align_start), align_size);
9049771f DW	382	err_pfn_remap:
9476df7d DW	383	err_radix:
	384	pgmap_radix_release(res);
	385	devres_free(page_map);
	386	return ERR_PTR(error);
41e94a85 CH	387	}
41e94a85 CH	388	EXPORT_SYMBOL(devm_memremap_pages);
4b94ffdc DW	389
	390	unsigned long vmem_altmap_offset(struct vmem_altmap *altmap)
	391	{
	392	/* number of pfns from base where pfn_to_page() is valid */
	393	return altmap->reserve + altmap->free;
	394	}
	395
	396	void vmem_altmap_free(struct vmem_altmap *altmap, unsigned long nr_pfns)
	397	{
	398	altmap->alloc -= nr_pfns;
	399	}
	400
4b94ffdc DW	401	struct vmem_altmap *to_vmem_altmap(unsigned long memmap_start)
	402	{
	403	/*
	404	* 'memmap_start' is the virtual address for the first "struct
	405	* page" in this range of the vmemmap array. In the case of
07061aab	406	* CONFIG_SPARSEMEM_VMEMMAP a page_to_pfn conversion is simple
4b94ffdc DW	407	* pointer arithmetic, so we can perform this to_vmem_altmap()
	408	* conversion without concern for the initialization state of
	409	* the struct page fields.
	410	*/
	411	struct page page = (struct page ) memmap_start;
	412	struct dev_pagemap *pgmap;
	413
	414	/*
07061aab	415	* Unconditionally retrieve a dev_pagemap associated with the
4b94ffdc DW	416	* given physical address, this is only for use in the
	417	* arch_{add\|remove}_memory() for setting up and tearing down
	418	* the memmap.
	419	*/
	420	rcu_read_lock();
	421	pgmap = find_dev_pagemap(__pfn_to_phys(page_to_pfn(page)));
	422	rcu_read_unlock();
	423
	424	return pgmap ? pgmap->altmap : NULL;
	425	}
41e94a85	426	#endif /* CONFIG_ZONE_DEVICE */