[mirror_ubuntu-eoan-kernel.git] / include / linux / memremap.h

/* SPDX-License-Identifier: GPL-2.0 */
#ifndef _LINUX_MEMREMAP_H_
#define _LINUX_MEMREMAP_H_
#include <linux/ioport.h>
#include <linux/percpu-refcount.h>

#include <asm/pgtable.h>

struct resource;
struct device;

/**
 * struct vmem_altmap - pre-allocated storage for vmemmap_populate
 * @base_pfn: base of the entire dev_pagemap mapping
 * @reserve: pages mapped, but reserved for driver use (relative to @base)
 * @free: free pages set aside in the mapping for memmap storage
 * @align: pages reserved to meet allocation alignments
 * @alloc: track pages consumed, private to vmemmap_populate()
 */
struct vmem_altmap {
	const unsigned long base_pfn;
	const unsigned long reserve;
	unsigned long free;
	unsigned long align;
	unsigned long alloc;
};

/*
 * Specialize ZONE_DEVICE memory into multiple types each having differents
 * usage.
 *
 * MEMORY_DEVICE_PRIVATE:
 * Device memory that is not directly addressable by the CPU: CPU can neither
 * read nor write private memory. In this case, we do still have struct pages
 * backing the device memory. Doing so simplifies the implementation, but it is
 * important to remember that there are certain points at which the struct page
 * must be treated as an opaque object, rather than a "normal" struct page.
 *
 * A more complete discussion of unaddressable memory may be found in
 * include/linux/hmm.h and Documentation/vm/hmm.rst.
 *
 * MEMORY_DEVICE_PUBLIC:
 * Device memory that is cache coherent from device and CPU point of view. This
 * is use on platform that have an advance system bus (like CAPI or CCIX). A
 * driver can hotplug the device memory using ZONE_DEVICE and with that memory
 * type. Any page of a process can be migrated to such memory. However no one
 * should be allow to pin such memory so that it can always be evicted.
 *
 * MEMORY_DEVICE_FS_DAX:
 * Host memory that has similar access semantics as System RAM i.e. DMA
 * coherent and supports page pinning. In support of coordinating page
 * pinning vs other operations MEMORY_DEVICE_FS_DAX arranges for a
 * wakeup event whenever a page is unpinned and becomes idle. This
 * wakeup is used to coordinate physical address space management (ex:
 * fs truncate/hole punch) vs pinned pages (ex: device dma).
 *
 * MEMORY_DEVICE_PCI_P2PDMA:
 * Device memory residing in a PCI BAR intended for use with Peer-to-Peer
 * transactions.
 */
enum memory_type {
	MEMORY_DEVICE_PRIVATE = 1,
	MEMORY_DEVICE_PUBLIC,
	MEMORY_DEVICE_FS_DAX,
	MEMORY_DEVICE_PCI_P2PDMA,
};

/*
 * For MEMORY_DEVICE_PRIVATE we use ZONE_DEVICE and extend it with two
 * callbacks:
 *   page_fault()
 *   page_free()
 *
 * Additional notes about MEMORY_DEVICE_PRIVATE may be found in
 * include/linux/hmm.h and Documentation/vm/hmm.rst. There is also a brief
 * explanation in include/linux/memory_hotplug.h.
 *
 * The page_fault() callback must migrate page back, from device memory to
 * system memory, so that the CPU can access it. This might fail for various
 * reasons (device issues,  device have been unplugged, ...). When such error
 * conditions happen, the page_fault() callback must return VM_FAULT_SIGBUS and
 * set the CPU page table entry to "poisoned".
 *
 * Note that because memory cgroup charges are transferred to the device memory,
 * this should never fail due to memory restrictions. However, allocation
 * of a regular system page might still fail because we are out of memory. If
 * that happens, the page_fault() callback must return VM_FAULT_OOM.
 *
 * The page_fault() callback can also try to migrate back multiple pages in one
 * chunk, as an optimization. It must, however, prioritize the faulting address
 * over all the others.
 *
 *
 * The page_free() callback is called once the page refcount reaches 1
 * (ZONE_DEVICE pages never reach 0 refcount unless there is a refcount bug.
 * This allows the device driver to implement its own memory management.)
 *
 * For MEMORY_DEVICE_PUBLIC only the page_free() callback matter.
 */
typedef int (*dev_page_fault_t)(struct vm_area_struct *vma,
				unsigned long addr,
				const struct page *page,
				unsigned int flags,
				pmd_t *pmdp);
typedef void (*dev_page_free_t)(struct page *page, void *data);

/**
 * struct dev_pagemap - metadata for ZONE_DEVICE mappings
 * @page_fault: callback when CPU fault on an unaddressable device page
 * @page_free: free page callback when page refcount reaches 1
 * @altmap: pre-allocated/reserved memory for vmemmap allocations
 * @res: physical address range covered by @ref
 * @ref: reference count that pins the devm_memremap_pages() mapping
 * @dev: host device of the mapping for debug
 * @data: private data pointer for page_free()
 * @type: memory type: see MEMORY_* in memory_hotplug.h
 */
struct dev_pagemap {
	dev_page_fault_t page_fault;
	dev_page_free_t page_free;
	struct vmem_altmap altmap;
	bool altmap_valid;
	struct resource res;
	struct percpu_ref *ref;
	struct device *dev;
	void *data;
	enum memory_type type;
	u64 pci_p2pdma_bus_offset;
};

#ifdef CONFIG_ZONE_DEVICE
void *devm_memremap_pages(struct device *dev, struct dev_pagemap *pgmap);
struct dev_pagemap *get_dev_pagemap(unsigned long pfn,
		struct dev_pagemap *pgmap);

unsigned long vmem_altmap_offset(struct vmem_altmap *altmap);
void vmem_altmap_free(struct vmem_altmap *altmap, unsigned long nr_pfns);
#else
static inline void *devm_memremap_pages(struct device *dev,
		struct dev_pagemap *pgmap)
{
	/*
	 * Fail attempts to call devm_memremap_pages() without
	 * ZONE_DEVICE support enabled, this requires callers to fall
	 * back to plain devm_memremap() based on config
	 */
	WARN_ON_ONCE(1);
	return ERR_PTR(-ENXIO);
}

static inline struct dev_pagemap *get_dev_pagemap(unsigned long pfn,
		struct dev_pagemap *pgmap)
{
	return NULL;
}

static inline unsigned long vmem_altmap_offset(struct vmem_altmap *altmap)
{
	return 0;
}

static inline void vmem_altmap_free(struct vmem_altmap *altmap,
		unsigned long nr_pfns)
{
}
#endif /* CONFIG_ZONE_DEVICE */

static inline void put_dev_pagemap(struct dev_pagemap *pgmap)
{
	if (pgmap)
		percpu_ref_put(pgmap->ref);
}
#endif /* _LINUX_MEMREMAP_H_ */
Commit	Line	Data
b2441318	1	/* SPDX-License-Identifier: GPL-2.0 */
9476df7d DW	2	#ifndef _LINUX_MEMREMAP_H_
9476df7d DW	3	#define _LINUX_MEMREMAP_H_
5c2c2587 DW	4	#include <linux/ioport.h>
5c2c2587 DW	5	#include <linux/percpu-refcount.h>
9476df7d	6
5042db43 JG	7	#include <asm/pgtable.h>
5042db43 JG	8
9476df7d DW	9	struct resource;
9476df7d DW	10	struct device;
4b94ffdc DW	11
	12	/**
	13	* struct vmem_altmap - pre-allocated storage for vmemmap_populate
	14	* @base_pfn: base of the entire dev_pagemap mapping
	15	* @reserve: pages mapped, but reserved for driver use (relative to @base)
	16	* @free: free pages set aside in the mapping for memmap storage
	17	* @align: pages reserved to meet allocation alignments
	18	* @alloc: track pages consumed, private to vmemmap_populate()
	19	*/
	20	struct vmem_altmap {
	21	const unsigned long base_pfn;
	22	const unsigned long reserve;
	23	unsigned long free;
	24	unsigned long align;
	25	unsigned long alloc;
	26	};
	27
5042db43 JG	28	/*
	29	* Specialize ZONE_DEVICE memory into multiple types each having differents
	30	* usage.
	31	*
5042db43 JG	32	* MEMORY_DEVICE_PRIVATE:
	33	* Device memory that is not directly addressable by the CPU: CPU can neither
	34	* read nor write private memory. In this case, we do still have struct pages
	35	* backing the device memory. Doing so simplifies the implementation, but it is
	36	* important to remember that there are certain points at which the struct page
	37	* must be treated as an opaque object, rather than a "normal" struct page.
	38	*
	39	* A more complete discussion of unaddressable memory may be found in
ad56b738	40	* include/linux/hmm.h and Documentation/vm/hmm.rst.
df6ad698 JG	41	*
	42	* MEMORY_DEVICE_PUBLIC:
	43	* Device memory that is cache coherent from device and CPU point of view. This
	44	* is use on platform that have an advance system bus (like CAPI or CCIX). A
	45	* driver can hotplug the device memory using ZONE_DEVICE and with that memory
	46	* type. Any page of a process can be migrated to such memory. However no one
	47	* should be allow to pin such memory so that it can always be evicted.
e7638488 DW	48	*
	49	* MEMORY_DEVICE_FS_DAX:
	50	* Host memory that has similar access semantics as System RAM i.e. DMA
	51	* coherent and supports page pinning. In support of coordinating page
	52	* pinning vs other operations MEMORY_DEVICE_FS_DAX arranges for a
	53	* wakeup event whenever a page is unpinned and becomes idle. This
	54	* wakeup is used to coordinate physical address space management (ex:
	55	* fs truncate/hole punch) vs pinned pages (ex: device dma).
52916982 LG	56	*
	57	* MEMORY_DEVICE_PCI_P2PDMA:
	58	* Device memory residing in a PCI BAR intended for use with Peer-to-Peer
	59	* transactions.
5042db43 JG	60	*/
5042db43 JG	61	enum memory_type {
e7638488	62	MEMORY_DEVICE_PRIVATE = 1,
df6ad698	63	MEMORY_DEVICE_PUBLIC,
e7638488	64	MEMORY_DEVICE_FS_DAX,
52916982	65	MEMORY_DEVICE_PCI_P2PDMA,
5042db43 JG	66	};
	67
	68	/*
	69	* For MEMORY_DEVICE_PRIVATE we use ZONE_DEVICE and extend it with two
	70	* callbacks:
	71	* page_fault()
	72	* page_free()
	73	*
	74	* Additional notes about MEMORY_DEVICE_PRIVATE may be found in
ad56b738	75	* include/linux/hmm.h and Documentation/vm/hmm.rst. There is also a brief
5042db43 JG	76	* explanation in include/linux/memory_hotplug.h.
	77	*
	78	* The page_fault() callback must migrate page back, from device memory to
	79	* system memory, so that the CPU can access it. This might fail for various
	80	* reasons (device issues, device have been unplugged, ...). When such error
	81	* conditions happen, the page_fault() callback must return VM_FAULT_SIGBUS and
	82	* set the CPU page table entry to "poisoned".
	83	*
	84	* Note that because memory cgroup charges are transferred to the device memory,
	85	* this should never fail due to memory restrictions. However, allocation
	86	* of a regular system page might still fail because we are out of memory. If
	87	* that happens, the page_fault() callback must return VM_FAULT_OOM.
	88	*
	89	* The page_fault() callback can also try to migrate back multiple pages in one
	90	* chunk, as an optimization. It must, however, prioritize the faulting address
	91	* over all the others.
	92	*
	93	*
	94	* The page_free() callback is called once the page refcount reaches 1
	95	* (ZONE_DEVICE pages never reach 0 refcount unless there is a refcount bug.
	96	* This allows the device driver to implement its own memory management.)
df6ad698 JG	97	*
df6ad698 JG	98	* For MEMORY_DEVICE_PUBLIC only the page_free() callback matter.
5042db43 JG	99	*/
	100	typedef int (dev_page_fault_t)(struct vm_area_struct vma,
	101	unsigned long addr,
	102	const struct page *page,
	103	unsigned int flags,
	104	pmd_t *pmdp);
	105	typedef void (dev_page_free_t)(struct page page, void *data);
	106
9476df7d DW	107	/**
9476df7d DW	108	* struct dev_pagemap - metadata for ZONE_DEVICE mappings
5042db43 JG	109	* @page_fault: callback when CPU fault on an unaddressable device page
5042db43 JG	110	* @page_free: free page callback when page refcount reaches 1
4b94ffdc	111	* @altmap: pre-allocated/reserved memory for vmemmap allocations
5c2c2587 DW	112	* @res: physical address range covered by @ref
5c2c2587 DW	113	* @ref: reference count that pins the devm_memremap_pages() mapping
9476df7d	114	* @dev: host device of the mapping for debug
5042db43 JG	115	* @data: private data pointer for page_free()
5042db43 JG	116	* @type: memory type: see MEMORY_* in memory_hotplug.h
9476df7d DW	117	*/
9476df7d DW	118	struct dev_pagemap {
5042db43 JG	119	dev_page_fault_t page_fault;
5042db43 JG	120	dev_page_free_t page_free;
e7744aa2 LG	121	struct vmem_altmap altmap;
	122	bool altmap_valid;
	123	struct resource res;
5c2c2587	124	struct percpu_ref *ref;
9476df7d	125	struct device *dev;
5042db43 JG	126	void *data;
5042db43 JG	127	enum memory_type type;
977196b8	128	u64 pci_p2pdma_bus_offset;
9476df7d DW	129	};
	130
	131	#ifdef CONFIG_ZONE_DEVICE
e8d51348	132	void devm_memremap_pages(struct device dev, struct dev_pagemap *pgmap);
0822acb8 CH	133	struct dev_pagemap *get_dev_pagemap(unsigned long pfn,
0822acb8 CH	134	struct dev_pagemap *pgmap);
7b2d55d2	135
8e37d00a CH	136	unsigned long vmem_altmap_offset(struct vmem_altmap *altmap);
8e37d00a CH	137	void vmem_altmap_free(struct vmem_altmap *altmap, unsigned long nr_pfns);
9476df7d DW	138	#else
9476df7d DW	139	static inline void devm_memremap_pages(struct device dev,
e8d51348	140	struct dev_pagemap *pgmap)
9476df7d DW	141	{
	142	/*
	143	* Fail attempts to call devm_memremap_pages() without
	144	* ZONE_DEVICE support enabled, this requires callers to fall
	145	* back to plain devm_memremap() based on config
	146	*/
	147	WARN_ON_ONCE(1);
	148	return ERR_PTR(-ENXIO);
	149	}
	150
0822acb8 CH	151	static inline struct dev_pagemap *get_dev_pagemap(unsigned long pfn,
0822acb8 CH	152	struct dev_pagemap *pgmap)
9476df7d DW	153	{
	154	return NULL;
	155	}
8e37d00a CH	156
	157	static inline unsigned long vmem_altmap_offset(struct vmem_altmap *altmap)
	158	{
	159	return 0;
	160	}
	161
	162	static inline void vmem_altmap_free(struct vmem_altmap *altmap,
	163	unsigned long nr_pfns)
	164	{
	165	}
	166	#endif /* CONFIG_ZONE_DEVICE */
7b2d55d2	167
5c2c2587 DW	168	static inline void put_dev_pagemap(struct dev_pagemap *pgmap)
	169	{
	170	if (pgmap)
	171	percpu_ref_put(pgmap->ref);
	172	}
9476df7d	173	#endif /* _LINUX_MEMREMAP_H_ */