[linux-block.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

static void *try_ram_remap(resource_size_t offset, size_t size)
{
	struct page *page = pfn_to_page(offset >> PAGE_SHIFT);

	/* In the simple case just return the existing linear address */
	if (!PageHighMem(page))
		return __va(offset);
	return NULL; /* fallback to ioremap_cache */
}

/**
 * memremap() - remap an iomem_resource as cacheable memory
 * @offset: iomem resource start address
 * @size: size of remap
 * @flags: either MEMREMAP_WB or MEMREMAP_WT
 *
 * 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.
 *
 * 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.
 */
void *memremap(resource_size_t offset, size_t size, unsigned long flags)
{
	int is_ram = region_intersects(offset, size, "System RAM");
	void *addr = 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) {
		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 = ioremap_cache(offset, size);
	}

	/*
	 * If we don't have a mapping yet and more request flags are
	 * pending 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) {
		WARN_ONCE(1, "memremap attempted on ram %pa size: %#lx\n",
				&offset, (unsigned long) size);
		return NULL;
	}

	if (!addr && (flags & MEMREMAP_WT)) {
		flags &= ~MEMREMAP_WT;
		addr = ioremap_wt(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(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 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);

pfn_t phys_to_pfn_t(dma_addr_t addr, unsigned long flags)
{
	return __pfn_to_pfn_t(addr >> PAGE_SHIFT, flags);
}
EXPORT_SYMBOL(phys_to_pfn_t);

#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;
};

static void pgmap_radix_release(struct resource *res)
{
	resource_size_t key;

	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 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;

	pgmap_radix_release(res);

	/* 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);
}

/* 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;
}

void *devm_memremap_pages(struct device *dev, struct resource *res)
{
	int is_ram = region_intersects(res->start, resource_size(res),
			"System RAM");
	resource_size_t key, align_start, align_size;
	struct page_map *page_map;
	int error, nid;

	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);

	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);

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

	page_map->pgmap.dev = dev;
	mutex_lock(&pgmap_lock);
	error = 0;
	for (key = res->start; key <= res->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();

	align_start = res->start & ~(SECTION_SIZE - 1);
	align_size = ALIGN(resource_size(res), SECTION_SIZE);
	error = arch_add_memory(nid, align_start, align_size, true);
	if (error)
		goto err_add_memory;

	devres_add(dev, page_map);
	return __va(res->start);

 err_add_memory:
 err_radix:
	pgmap_radix_release(res);
	devres_free(page_map);
	return ERR_PTR(error);
}
EXPORT_SYMBOL(devm_memremap_pages);
#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
182475b7 DW	30	static void *try_ram_remap(resource_size_t offset, size_t size)
	31	{
	32	struct page *page = pfn_to_page(offset >> PAGE_SHIFT);
	33
	34	/* In the simple case just return the existing linear address */
	35	if (!PageHighMem(page))
	36	return __va(offset);
	37	return NULL; /* fallback to ioremap_cache */
	38	}
	39
92281dee DW	40	/**
	41	* memremap() - remap an iomem_resource as cacheable memory
	42	* @offset: iomem resource start address
	43	* @size: size of remap
	44	* @flags: either MEMREMAP_WB or MEMREMAP_WT
	45	*
	46	* memremap() is "ioremap" for cases where it is known that the resource
	47	* being mapped does not have i/o side effects and the __iomem
	48	* annotation is not applicable.
	49	*
	50	* MEMREMAP_WB - matches the default mapping for "System RAM" on
	51	* the architecture. This is usually a read-allocate write-back cache.
	52	* Morever, if MEMREMAP_WB is specified and the requested remap region is RAM
	53	* memremap() will bypass establishing a new mapping and instead return
	54	* a pointer into the direct map.
	55	*
	56	* MEMREMAP_WT - establish a mapping whereby writes either bypass the
	57	* cache or are written through to memory and never exist in a
	58	* cache-dirty state with respect to program visibility. Attempts to
	59	* map "System RAM" with this mapping type will fail.
	60	*/
	61	void *memremap(resource_size_t offset, size_t size, unsigned long flags)
	62	{
	63	int is_ram = region_intersects(offset, size, "System RAM");
	64	void *addr = NULL;
	65
	66	if (is_ram == REGION_MIXED) {
	67	WARN_ONCE(1, "memremap attempted on mixed range %pa size: %#lx\n",
	68	&offset, (unsigned long) size);
	69	return NULL;
	70	}
	71
	72	/* Try all mapping types requested until one returns non-NULL */
	73	if (flags & MEMREMAP_WB) {
	74	flags &= ~MEMREMAP_WB;
	75	/*
	76	* MEMREMAP_WB is special in that it can be satisifed
	77	* from the direct map. Some archs depend on the
	78	* capability of memremap() to autodetect cases where
	79	* the requested range is potentially in "System RAM"
	80	*/
	81	if (is_ram == REGION_INTERSECTS)
182475b7 DW	82	addr = try_ram_remap(offset, size);
182475b7 DW	83	if (!addr)
92281dee DW	84	addr = ioremap_cache(offset, size);
	85	}
	86
	87	/*
	88	* If we don't have a mapping yet and more request flags are
	89	* pending then we will be attempting to establish a new virtual
	90	* address mapping. Enforce that this mapping is not aliasing
	91	* "System RAM"
	92	*/
	93	if (!addr && is_ram == REGION_INTERSECTS && flags) {
	94	WARN_ONCE(1, "memremap attempted on ram %pa size: %#lx\n",
	95	&offset, (unsigned long) size);
	96	return NULL;
	97	}
	98
	99	if (!addr && (flags & MEMREMAP_WT)) {
	100	flags &= ~MEMREMAP_WT;
	101	addr = ioremap_wt(offset, size);
	102	}
	103
	104	return addr;
	105	}
	106	EXPORT_SYMBOL(memremap);
	107
	108	void memunmap(void *addr)
	109	{
	110	if (is_vmalloc_addr(addr))
	111	iounmap((void __iomem *) addr);
	112	}
	113	EXPORT_SYMBOL(memunmap);
7d3dcf26 CH	114
	115	static void devm_memremap_release(struct device dev, void res)
	116	{
	117	memunmap(res);
	118	}
	119
	120	static int devm_memremap_match(struct device dev, void res, void *match_data)
	121	{
	122	return (void *)res == match_data;
	123	}
	124
	125	void devm_memremap(struct device dev, resource_size_t offset,
	126	size_t size, unsigned long flags)
	127	{
	128	void *ptr, addr;
	129
538ea4aa DW	130	ptr = devres_alloc_node(devm_memremap_release, sizeof(*ptr), GFP_KERNEL,
538ea4aa DW	131	dev_to_node(dev));
7d3dcf26	132	if (!ptr)
b36f4761	133	return ERR_PTR(-ENOMEM);
7d3dcf26 CH	134
	135	addr = memremap(offset, size, flags);
	136	if (addr) {
	137	*ptr = addr;
	138	devres_add(dev, ptr);
	139	} else
	140	devres_free(ptr);
	141
	142	return addr;
	143	}
	144	EXPORT_SYMBOL(devm_memremap);
	145
	146	void devm_memunmap(struct device dev, void addr)
	147	{
d741314f DW	148	WARN_ON(devres_release(dev, devm_memremap_release,
d741314f DW	149	devm_memremap_match, addr));
7d3dcf26 CH	150	}
7d3dcf26 CH	151	EXPORT_SYMBOL(devm_memunmap);
41e94a85	152
34c0fd54 DW	153	pfn_t phys_to_pfn_t(dma_addr_t addr, unsigned long flags)
	154	{
	155	return __pfn_to_pfn_t(addr >> PAGE_SHIFT, flags);
	156	}
	157	EXPORT_SYMBOL(phys_to_pfn_t);
	158
41e94a85	159	#ifdef CONFIG_ZONE_DEVICE
9476df7d DW	160	static DEFINE_MUTEX(pgmap_lock);
	161	static RADIX_TREE(pgmap_radix, GFP_KERNEL);
	162	#define SECTION_MASK ~((1UL << PA_SECTION_SHIFT) - 1)
	163	#define SECTION_SIZE (1UL << PA_SECTION_SHIFT)
	164
41e94a85 CH	165	struct page_map {
41e94a85 CH	166	struct resource res;
9476df7d DW	167	struct percpu_ref *ref;
9476df7d DW	168	struct dev_pagemap pgmap;
41e94a85 CH	169	};
41e94a85 CH	170
9476df7d DW	171	static void pgmap_radix_release(struct resource *res)
	172	{
	173	resource_size_t key;
	174
	175	mutex_lock(&pgmap_lock);
	176	for (key = res->start; key <= res->end; key += SECTION_SIZE)
	177	radix_tree_delete(&pgmap_radix, key >> PA_SECTION_SHIFT);
	178	mutex_unlock(&pgmap_lock);
	179	}
	180
	181	static void devm_memremap_pages_release(struct device dev, void data)
41e94a85	182	{
9476df7d DW	183	struct page_map *page_map = data;
	184	struct resource *res = &page_map->res;
	185	resource_size_t align_start, align_size;
	186
	187	pgmap_radix_release(res);
41e94a85 CH	188
41e94a85 CH	189	/* pages are dead and unused, undo the arch mapping */
9476df7d DW	190	align_start = res->start & ~(SECTION_SIZE - 1);
	191	align_size = ALIGN(resource_size(res), SECTION_SIZE);
	192	arch_remove_memory(align_start, align_size);
	193	}
	194
	195	/* assumes rcu_read_lock() held at entry */
	196	struct dev_pagemap *find_dev_pagemap(resource_size_t phys)
	197	{
	198	struct page_map *page_map;
	199
	200	WARN_ON_ONCE(!rcu_read_lock_held());
	201
	202	page_map = radix_tree_lookup(&pgmap_radix, phys >> PA_SECTION_SHIFT);
	203	return page_map ? &page_map->pgmap : NULL;
41e94a85 CH	204	}
	205
	206	void devm_memremap_pages(struct device dev, struct resource *res)
	207	{
	208	int is_ram = region_intersects(res->start, resource_size(res),
	209	"System RAM");
9476df7d	210	resource_size_t key, align_start, align_size;
41e94a85 CH	211	struct page_map *page_map;
	212	int error, nid;
	213
	214	if (is_ram == REGION_MIXED) {
	215	WARN_ONCE(1, "%s attempted on mixed region %pr\n",
	216	__func__, res);
	217	return ERR_PTR(-ENXIO);
	218	}
	219
	220	if (is_ram == REGION_INTERSECTS)
	221	return __va(res->start);
	222
538ea4aa DW	223	page_map = devres_alloc_node(devm_memremap_pages_release,
538ea4aa DW	224	sizeof(*page_map), GFP_KERNEL, dev_to_node(dev));
41e94a85 CH	225	if (!page_map)
	226	return ERR_PTR(-ENOMEM);
	227
	228	memcpy(&page_map->res, res, sizeof(*res));
	229
9476df7d DW	230	page_map->pgmap.dev = dev;
	231	mutex_lock(&pgmap_lock);
	232	error = 0;
	233	for (key = res->start; key <= res->end; key += SECTION_SIZE) {
	234	struct dev_pagemap *dup;
	235
	236	rcu_read_lock();
	237	dup = find_dev_pagemap(key);
	238	rcu_read_unlock();
	239	if (dup) {
	240	dev_err(dev, "%s: %pr collides with mapping for %s\n",
	241	__func__, res, dev_name(dup->dev));
	242	error = -EBUSY;
	243	break;
	244	}
	245	error = radix_tree_insert(&pgmap_radix, key >> PA_SECTION_SHIFT,
	246	page_map);
	247	if (error) {
	248	dev_err(dev, "%s: failed: %d\n", __func__, error);
	249	break;
	250	}
	251	}
	252	mutex_unlock(&pgmap_lock);
	253	if (error)
	254	goto err_radix;
	255
41e94a85 CH	256	nid = dev_to_node(dev);
41e94a85 CH	257	if (nid < 0)
7eff93b7	258	nid = numa_mem_id();
41e94a85	259
9476df7d DW	260	align_start = res->start & ~(SECTION_SIZE - 1);
	261	align_size = ALIGN(resource_size(res), SECTION_SIZE);
	262	error = arch_add_memory(nid, align_start, align_size, true);
	263	if (error)
	264	goto err_add_memory;
41e94a85 CH	265
	266	devres_add(dev, page_map);
	267	return __va(res->start);
9476df7d DW	268
	269	err_add_memory:
	270	err_radix:
	271	pgmap_radix_release(res);
	272	devres_free(page_map);
	273	return ERR_PTR(error);
41e94a85 CH	274	}
	275	EXPORT_SYMBOL(devm_memremap_pages);
	276	#endif /* CONFIG_ZONE_DEVICE */