[linux-block.git] / drivers / dax / device.c

// SPDX-License-Identifier: GPL-2.0
/* Copyright(c) 2016-2018 Intel Corporation. All rights reserved. */
#include <linux/memremap.h>
#include <linux/pagemap.h>
#include <linux/module.h>
#include <linux/device.h>
#include <linux/pfn_t.h>
#include <linux/cdev.h>
#include <linux/slab.h>
#include <linux/dax.h>
#include <linux/fs.h>
#include <linux/mm.h>
#include <linux/mman.h>
#include "dax-private.h"
#include "bus.h"

static int check_vma(struct dev_dax *dev_dax, struct vm_area_struct *vma,
		const char *func)
{
	struct device *dev = &dev_dax->dev;
	unsigned long mask;

	if (!dax_alive(dev_dax->dax_dev))
		return -ENXIO;

	/* prevent private mappings from being established */
	if ((vma->vm_flags & VM_MAYSHARE) != VM_MAYSHARE) {
		dev_info_ratelimited(dev,
				"%s: %s: fail, attempted private mapping\n",
				current->comm, func);
		return -EINVAL;
	}

	mask = dev_dax->align - 1;
	if (vma->vm_start & mask || vma->vm_end & mask) {
		dev_info_ratelimited(dev,
				"%s: %s: fail, unaligned vma (%#lx - %#lx, %#lx)\n",
				current->comm, func, vma->vm_start, vma->vm_end,
				mask);
		return -EINVAL;
	}

	if (!vma_is_dax(vma)) {
		dev_info_ratelimited(dev,
				"%s: %s: fail, vma is not DAX capable\n",
				current->comm, func);
		return -EINVAL;
	}

	return 0;
}

/* see "strong" declaration in tools/testing/nvdimm/dax-dev.c */
__weak phys_addr_t dax_pgoff_to_phys(struct dev_dax *dev_dax, pgoff_t pgoff,
		unsigned long size)
{
	int i;

	for (i = 0; i < dev_dax->nr_range; i++) {
		struct dev_dax_range *dax_range = &dev_dax->ranges[i];
		struct range *range = &dax_range->range;
		unsigned long long pgoff_end;
		phys_addr_t phys;

		pgoff_end = dax_range->pgoff + PHYS_PFN(range_len(range)) - 1;
		if (pgoff < dax_range->pgoff || pgoff > pgoff_end)
			continue;
		phys = PFN_PHYS(pgoff - dax_range->pgoff) + range->start;
		if (phys + size - 1 <= range->end)
			return phys;
		break;
	}
	return -1;
}

static void dax_set_mapping(struct vm_fault *vmf, pfn_t pfn,
			      unsigned long fault_size)
{
	unsigned long i, nr_pages = fault_size / PAGE_SIZE;
	struct file *filp = vmf->vma->vm_file;
	struct dev_dax *dev_dax = filp->private_data;
	pgoff_t pgoff;

	/* mapping is only set on the head */
	if (dev_dax->pgmap->vmemmap_shift)
		nr_pages = 1;

	pgoff = linear_page_index(vmf->vma,
			ALIGN(vmf->address, fault_size));

	for (i = 0; i < nr_pages; i++) {
		struct page *page = pfn_to_page(pfn_t_to_pfn(pfn) + i);

		page = compound_head(page);
		if (page->mapping)
			continue;

		page->mapping = filp->f_mapping;
		page->index = pgoff + i;
	}
}

static vm_fault_t __dev_dax_pte_fault(struct dev_dax *dev_dax,
				struct vm_fault *vmf)
{
	struct device *dev = &dev_dax->dev;
	phys_addr_t phys;
	pfn_t pfn;
	unsigned int fault_size = PAGE_SIZE;

	if (check_vma(dev_dax, vmf->vma, __func__))
		return VM_FAULT_SIGBUS;

	if (dev_dax->align > PAGE_SIZE) {
		dev_dbg(dev, "alignment (%#x) > fault size (%#x)\n",
			dev_dax->align, fault_size);
		return VM_FAULT_SIGBUS;
	}

	if (fault_size != dev_dax->align)
		return VM_FAULT_SIGBUS;

	phys = dax_pgoff_to_phys(dev_dax, vmf->pgoff, PAGE_SIZE);
	if (phys == -1) {
		dev_dbg(dev, "pgoff_to_phys(%#lx) failed\n", vmf->pgoff);
		return VM_FAULT_SIGBUS;
	}

	pfn = phys_to_pfn_t(phys, PFN_DEV|PFN_MAP);

	dax_set_mapping(vmf, pfn, fault_size);

	return vmf_insert_mixed(vmf->vma, vmf->address, pfn);
}

static vm_fault_t __dev_dax_pmd_fault(struct dev_dax *dev_dax,
				struct vm_fault *vmf)
{
	unsigned long pmd_addr = vmf->address & PMD_MASK;
	struct device *dev = &dev_dax->dev;
	phys_addr_t phys;
	pgoff_t pgoff;
	pfn_t pfn;
	unsigned int fault_size = PMD_SIZE;

	if (check_vma(dev_dax, vmf->vma, __func__))
		return VM_FAULT_SIGBUS;

	if (dev_dax->align > PMD_SIZE) {
		dev_dbg(dev, "alignment (%#x) > fault size (%#x)\n",
			dev_dax->align, fault_size);
		return VM_FAULT_SIGBUS;
	}

	if (fault_size < dev_dax->align)
		return VM_FAULT_SIGBUS;
	else if (fault_size > dev_dax->align)
		return VM_FAULT_FALLBACK;

	/* if we are outside of the VMA */
	if (pmd_addr < vmf->vma->vm_start ||
			(pmd_addr + PMD_SIZE) > vmf->vma->vm_end)
		return VM_FAULT_SIGBUS;

	pgoff = linear_page_index(vmf->vma, pmd_addr);
	phys = dax_pgoff_to_phys(dev_dax, pgoff, PMD_SIZE);
	if (phys == -1) {
		dev_dbg(dev, "pgoff_to_phys(%#lx) failed\n", pgoff);
		return VM_FAULT_SIGBUS;
	}

	pfn = phys_to_pfn_t(phys, PFN_DEV|PFN_MAP);

	dax_set_mapping(vmf, pfn, fault_size);

	return vmf_insert_pfn_pmd(vmf, pfn, vmf->flags & FAULT_FLAG_WRITE);
}

#ifdef CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD
static vm_fault_t __dev_dax_pud_fault(struct dev_dax *dev_dax,
				struct vm_fault *vmf)
{
	unsigned long pud_addr = vmf->address & PUD_MASK;
	struct device *dev = &dev_dax->dev;
	phys_addr_t phys;
	pgoff_t pgoff;
	pfn_t pfn;
	unsigned int fault_size = PUD_SIZE;


	if (check_vma(dev_dax, vmf->vma, __func__))
		return VM_FAULT_SIGBUS;

	if (dev_dax->align > PUD_SIZE) {
		dev_dbg(dev, "alignment (%#x) > fault size (%#x)\n",
			dev_dax->align, fault_size);
		return VM_FAULT_SIGBUS;
	}

	if (fault_size < dev_dax->align)
		return VM_FAULT_SIGBUS;
	else if (fault_size > dev_dax->align)
		return VM_FAULT_FALLBACK;

	/* if we are outside of the VMA */
	if (pud_addr < vmf->vma->vm_start ||
			(pud_addr + PUD_SIZE) > vmf->vma->vm_end)
		return VM_FAULT_SIGBUS;

	pgoff = linear_page_index(vmf->vma, pud_addr);
	phys = dax_pgoff_to_phys(dev_dax, pgoff, PUD_SIZE);
	if (phys == -1) {
		dev_dbg(dev, "pgoff_to_phys(%#lx) failed\n", pgoff);
		return VM_FAULT_SIGBUS;
	}

	pfn = phys_to_pfn_t(phys, PFN_DEV|PFN_MAP);

	dax_set_mapping(vmf, pfn, fault_size);

	return vmf_insert_pfn_pud(vmf, pfn, vmf->flags & FAULT_FLAG_WRITE);
}
#else
static vm_fault_t __dev_dax_pud_fault(struct dev_dax *dev_dax,
				struct vm_fault *vmf)
{
	return VM_FAULT_FALLBACK;
}
#endif /* !CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD */

static vm_fault_t dev_dax_huge_fault(struct vm_fault *vmf,
		enum page_entry_size pe_size)
{
	struct file *filp = vmf->vma->vm_file;
	vm_fault_t rc = VM_FAULT_SIGBUS;
	int id;
	struct dev_dax *dev_dax = filp->private_data;

	dev_dbg(&dev_dax->dev, "%s: %s (%#lx - %#lx) size = %d\n", current->comm,
			(vmf->flags & FAULT_FLAG_WRITE) ? "write" : "read",
			vmf->vma->vm_start, vmf->vma->vm_end, pe_size);

	id = dax_read_lock();
	switch (pe_size) {
	case PE_SIZE_PTE:
		rc = __dev_dax_pte_fault(dev_dax, vmf);
		break;
	case PE_SIZE_PMD:
		rc = __dev_dax_pmd_fault(dev_dax, vmf);
		break;
	case PE_SIZE_PUD:
		rc = __dev_dax_pud_fault(dev_dax, vmf);
		break;
	default:
		rc = VM_FAULT_SIGBUS;
	}

	dax_read_unlock(id);

	return rc;
}

static vm_fault_t dev_dax_fault(struct vm_fault *vmf)
{
	return dev_dax_huge_fault(vmf, PE_SIZE_PTE);
}

static int dev_dax_may_split(struct vm_area_struct *vma, unsigned long addr)
{
	struct file *filp = vma->vm_file;
	struct dev_dax *dev_dax = filp->private_data;

	if (!IS_ALIGNED(addr, dev_dax->align))
		return -EINVAL;
	return 0;
}

static unsigned long dev_dax_pagesize(struct vm_area_struct *vma)
{
	struct file *filp = vma->vm_file;
	struct dev_dax *dev_dax = filp->private_data;

	return dev_dax->align;
}

static const struct vm_operations_struct dax_vm_ops = {
	.fault = dev_dax_fault,
	.huge_fault = dev_dax_huge_fault,
	.may_split = dev_dax_may_split,
	.pagesize = dev_dax_pagesize,
};

static int dax_mmap(struct file *filp, struct vm_area_struct *vma)
{
	struct dev_dax *dev_dax = filp->private_data;
	int rc, id;

	dev_dbg(&dev_dax->dev, "trace\n");

	/*
	 * We lock to check dax_dev liveness and will re-check at
	 * fault time.
	 */
	id = dax_read_lock();
	rc = check_vma(dev_dax, vma, __func__);
	dax_read_unlock(id);
	if (rc)
		return rc;

	vma->vm_ops = &dax_vm_ops;
	vma->vm_flags |= VM_HUGEPAGE;
	return 0;
}

/* return an unmapped area aligned to the dax region specified alignment */
static unsigned long dax_get_unmapped_area(struct file *filp,
		unsigned long addr, unsigned long len, unsigned long pgoff,
		unsigned long flags)
{
	unsigned long off, off_end, off_align, len_align, addr_align, align;
	struct dev_dax *dev_dax = filp ? filp->private_data : NULL;

	if (!dev_dax || addr)
		goto out;

	align = dev_dax->align;
	off = pgoff << PAGE_SHIFT;
	off_end = off + len;
	off_align = round_up(off, align);

	if ((off_end <= off_align) || ((off_end - off_align) < align))
		goto out;

	len_align = len + align;
	if ((off + len_align) < off)
		goto out;

	addr_align = current->mm->get_unmapped_area(filp, addr, len_align,
			pgoff, flags);
	if (!IS_ERR_VALUE(addr_align)) {
		addr_align += (off - addr_align) & (align - 1);
		return addr_align;
	}
 out:
	return current->mm->get_unmapped_area(filp, addr, len, pgoff, flags);
}

static const struct address_space_operations dev_dax_aops = {
	.dirty_folio	= noop_dirty_folio,
};

static int dax_open(struct inode *inode, struct file *filp)
{
	struct dax_device *dax_dev = inode_dax(inode);
	struct inode *__dax_inode = dax_inode(dax_dev);
	struct dev_dax *dev_dax = dax_get_private(dax_dev);

	dev_dbg(&dev_dax->dev, "trace\n");
	inode->i_mapping = __dax_inode->i_mapping;
	inode->i_mapping->host = __dax_inode;
	inode->i_mapping->a_ops = &dev_dax_aops;
	filp->f_mapping = inode->i_mapping;
	filp->f_wb_err = filemap_sample_wb_err(filp->f_mapping);
	filp->f_sb_err = file_sample_sb_err(filp);
	filp->private_data = dev_dax;
	inode->i_flags = S_DAX;

	return 0;
}

static int dax_release(struct inode *inode, struct file *filp)
{
	struct dev_dax *dev_dax = filp->private_data;

	dev_dbg(&dev_dax->dev, "trace\n");
	return 0;
}

static const struct file_operations dax_fops = {
	.llseek = noop_llseek,
	.owner = THIS_MODULE,
	.open = dax_open,
	.release = dax_release,
	.get_unmapped_area = dax_get_unmapped_area,
	.mmap = dax_mmap,
	.mmap_supported_flags = MAP_SYNC,
};

static void dev_dax_cdev_del(void *cdev)
{
	cdev_del(cdev);
}

static void dev_dax_kill(void *dev_dax)
{
	kill_dev_dax(dev_dax);
}

int dev_dax_probe(struct dev_dax *dev_dax)
{
	struct dax_device *dax_dev = dev_dax->dax_dev;
	struct device *dev = &dev_dax->dev;
	struct dev_pagemap *pgmap;
	struct inode *inode;
	struct cdev *cdev;
	void *addr;
	int rc, i;

	if (static_dev_dax(dev_dax))  {
		if (dev_dax->nr_range > 1) {
			dev_warn(dev,
				"static pgmap / multi-range device conflict\n");
			return -EINVAL;
		}

		pgmap = dev_dax->pgmap;
	} else {
		if (dev_dax->pgmap) {
			dev_warn(dev,
				 "dynamic-dax with pre-populated page map\n");
			return -EINVAL;
		}

		pgmap = devm_kzalloc(dev,
                       struct_size(pgmap, ranges, dev_dax->nr_range - 1),
                       GFP_KERNEL);
		if (!pgmap)
			return -ENOMEM;

		pgmap->nr_range = dev_dax->nr_range;
		dev_dax->pgmap = pgmap;

		for (i = 0; i < dev_dax->nr_range; i++) {
			struct range *range = &dev_dax->ranges[i].range;
			pgmap->ranges[i] = *range;
		}
	}

	for (i = 0; i < dev_dax->nr_range; i++) {
		struct range *range = &dev_dax->ranges[i].range;

		if (!devm_request_mem_region(dev, range->start,
					range_len(range), dev_name(dev))) {
			dev_warn(dev, "mapping%d: %#llx-%#llx could not reserve range\n",
					i, range->start, range->end);
			return -EBUSY;
		}
	}

	pgmap->type = MEMORY_DEVICE_GENERIC;
	if (dev_dax->align > PAGE_SIZE)
		pgmap->vmemmap_shift =
			order_base_2(dev_dax->align >> PAGE_SHIFT);
	addr = devm_memremap_pages(dev, pgmap);
	if (IS_ERR(addr))
		return PTR_ERR(addr);

	inode = dax_inode(dax_dev);
	cdev = inode->i_cdev;
	cdev_init(cdev, &dax_fops);
	cdev->owner = dev->driver->owner;
	cdev_set_parent(cdev, &dev->kobj);
	rc = cdev_add(cdev, dev->devt, 1);
	if (rc)
		return rc;

	rc = devm_add_action_or_reset(dev, dev_dax_cdev_del, cdev);
	if (rc)
		return rc;

	run_dax(dax_dev);
	return devm_add_action_or_reset(dev, dev_dax_kill, dev_dax);
}
EXPORT_SYMBOL_GPL(dev_dax_probe);

static struct dax_device_driver device_dax_driver = {
	.probe = dev_dax_probe,
	/* all probe actions are unwound by devm, so .remove isn't necessary */
	.match_always = 1,
};

static int __init dax_init(void)
{
	return dax_driver_register(&device_dax_driver);
}

static void __exit dax_exit(void)
{
	dax_driver_unregister(&device_dax_driver);
}

MODULE_AUTHOR("Intel Corporation");
MODULE_LICENSE("GPL v2");
module_init(dax_init);
module_exit(dax_exit);
MODULE_ALIAS_DAX_DEVICE(0);
Commit	Line	Data
51cf784c DW	1	// SPDX-License-Identifier: GPL-2.0
51cf784c DW	2	/* Copyright(c) 2016-2018 Intel Corporation. All rights reserved. */
89ec9f2c	3	#include <linux/memremap.h>
ab68f262 DW	4	#include <linux/pagemap.h>
	5	#include <linux/module.h>
	6	#include <linux/device.h>
	7	#include <linux/pfn_t.h>
ba09c01d	8	#include <linux/cdev.h>
ab68f262 DW	9	#include <linux/slab.h>
	10	#include <linux/dax.h>
	11	#include <linux/fs.h>
	12	#include <linux/mm.h>
ef842302	13	#include <linux/mman.h>
efebc711	14	#include "dax-private.h"
51cf784c	15	#include "bus.h"
ab68f262	16
5f0694b3	17	static int check_vma(struct dev_dax dev_dax, struct vm_area_struct vma,
dee41079 DW	18	const char *func)
dee41079 DW	19	{
5f0694b3	20	struct device *dev = &dev_dax->dev;
dee41079 DW	21	unsigned long mask;
dee41079 DW	22
7b6be844	23	if (!dax_alive(dev_dax->dax_dev))
dee41079 DW	24	return -ENXIO;
dee41079 DW	25
4cb19355	26	/* prevent private mappings from being established */
325896ff	27	if ((vma->vm_flags & VM_MAYSHARE) != VM_MAYSHARE) {
5a14e91d JM	28	dev_info_ratelimited(dev,
5a14e91d JM	29	"%s: %s: fail, attempted private mapping\n",
dee41079 DW	30	current->comm, func);
	31	return -EINVAL;
	32	}
	33
33cf94d7	34	mask = dev_dax->align - 1;
dee41079	35	if (vma->vm_start & mask \|\| vma->vm_end & mask) {
5a14e91d JM	36	dev_info_ratelimited(dev,
5a14e91d JM	37	"%s: %s: fail, unaligned vma (%#lx - %#lx, %#lx)\n",
dee41079 DW	38	current->comm, func, vma->vm_start, vma->vm_end,
	39	mask);
	40	return -EINVAL;
	41	}
	42
dee41079	43	if (!vma_is_dax(vma)) {
5a14e91d JM	44	dev_info_ratelimited(dev,
5a14e91d JM	45	"%s: %s: fail, vma is not DAX capable\n",
dee41079 DW	46	current->comm, func);
	47	return -EINVAL;
	48	}
	49
	50	return 0;
	51	}
	52
efebc711	53	/* see "strong" declaration in tools/testing/nvdimm/dax-dev.c */
73616367	54	__weak phys_addr_t dax_pgoff_to_phys(struct dev_dax *dev_dax, pgoff_t pgoff,
dee41079 DW	55	unsigned long size)
dee41079 DW	56	{
60e93dc0 DW	57	int i;
	58
	59	for (i = 0; i < dev_dax->nr_range; i++) {
	60	struct dev_dax_range *dax_range = &dev_dax->ranges[i];
	61	struct range *range = &dax_range->range;
	62	unsigned long long pgoff_end;
	63	phys_addr_t phys;
	64
	65	pgoff_end = dax_range->pgoff + PHYS_PFN(range_len(range)) - 1;
	66	if (pgoff < dax_range->pgoff \|\| pgoff > pgoff_end)
	67	continue;
	68	phys = PFN_PHYS(pgoff - dax_range->pgoff) + range->start;
f5516ec5	69	if (phys + size - 1 <= range->end)
dee41079	70	return phys;
60e93dc0	71	break;
dee41079	72	}
dee41079 DW	73	return -1;
	74	}
	75
a0fb038e JM	76	static void dax_set_mapping(struct vm_fault *vmf, pfn_t pfn,
	77	unsigned long fault_size)
	78	{
	79	unsigned long i, nr_pages = fault_size / PAGE_SIZE;
	80	struct file *filp = vmf->vma->vm_file;
14606001	81	struct dev_dax *dev_dax = filp->private_data;
a0fb038e JM	82	pgoff_t pgoff;
a0fb038e JM	83
14606001 JM	84	/* mapping is only set on the head */
	85	if (dev_dax->pgmap->vmemmap_shift)
	86	nr_pages = 1;
	87
a0fb038e JM	88	pgoff = linear_page_index(vmf->vma,
	89	ALIGN(vmf->address, fault_size));
	90
	91	for (i = 0; i < nr_pages; i++) {
	92	struct page *page = pfn_to_page(pfn_t_to_pfn(pfn) + i);
	93
14606001	94	page = compound_head(page);
a0fb038e JM	95	if (page->mapping)
	96	continue;
	97
	98	page->mapping = filp->f_mapping;
	99	page->index = pgoff + i;
	100	}
	101	}
	102
226ab561	103	static vm_fault_t __dev_dax_pte_fault(struct dev_dax *dev_dax,
6ec228b6	104	struct vm_fault *vmf)
dee41079	105	{
5f0694b3	106	struct device *dev = &dev_dax->dev;
dee41079	107	phys_addr_t phys;
6ec228b6	108	pfn_t pfn;
0134ed4f	109	unsigned int fault_size = PAGE_SIZE;
dee41079	110
5f0694b3	111	if (check_vma(dev_dax, vmf->vma, __func__))
dee41079 DW	112	return VM_FAULT_SIGBUS;
dee41079 DW	113
33cf94d7	114	if (dev_dax->align > PAGE_SIZE) {
6daaca52	115	dev_dbg(dev, "alignment (%#x) > fault size (%#x)\n",
33cf94d7	116	dev_dax->align, fault_size);
dee41079 DW	117	return VM_FAULT_SIGBUS;
	118	}
	119
33cf94d7	120	if (fault_size != dev_dax->align)
0134ed4f DJ	121	return VM_FAULT_SIGBUS;
0134ed4f DJ	122
73616367	123	phys = dax_pgoff_to_phys(dev_dax, vmf->pgoff, PAGE_SIZE);
dee41079	124	if (phys == -1) {
6daaca52	125	dev_dbg(dev, "pgoff_to_phys(%#lx) failed\n", vmf->pgoff);
dee41079 DW	126	return VM_FAULT_SIGBUS;
	127	}
	128
6ec228b6	129	pfn = phys_to_pfn_t(phys, PFN_DEV\|PFN_MAP);
dee41079	130
6ec228b6	131	dax_set_mapping(vmf, pfn, fault_size);
0e7325f0	132
6ec228b6	133	return vmf_insert_mixed(vmf->vma, vmf->address, pfn);
dee41079 DW	134	}
dee41079 DW	135
226ab561	136	static vm_fault_t __dev_dax_pmd_fault(struct dev_dax *dev_dax,
6ec228b6	137	struct vm_fault *vmf)
dee41079	138	{
d8a849e1	139	unsigned long pmd_addr = vmf->address & PMD_MASK;
5f0694b3	140	struct device *dev = &dev_dax->dev;
dee41079 DW	141	phys_addr_t phys;
dee41079 DW	142	pgoff_t pgoff;
6ec228b6	143	pfn_t pfn;
0134ed4f	144	unsigned int fault_size = PMD_SIZE;
dee41079	145
5f0694b3	146	if (check_vma(dev_dax, vmf->vma, __func__))
dee41079 DW	147	return VM_FAULT_SIGBUS;
dee41079 DW	148
33cf94d7	149	if (dev_dax->align > PMD_SIZE) {
6daaca52	150	dev_dbg(dev, "alignment (%#x) > fault size (%#x)\n",
33cf94d7	151	dev_dax->align, fault_size);
dee41079 DW	152	return VM_FAULT_SIGBUS;
	153	}
	154
33cf94d7	155	if (fault_size < dev_dax->align)
0134ed4f	156	return VM_FAULT_SIGBUS;
33cf94d7	157	else if (fault_size > dev_dax->align)
0134ed4f DJ	158	return VM_FAULT_FALLBACK;
	159
	160	/* if we are outside of the VMA */
	161	if (pmd_addr < vmf->vma->vm_start \|\|
	162	(pmd_addr + PMD_SIZE) > vmf->vma->vm_end)
	163	return VM_FAULT_SIGBUS;
	164
f4200391	165	pgoff = linear_page_index(vmf->vma, pmd_addr);
73616367	166	phys = dax_pgoff_to_phys(dev_dax, pgoff, PMD_SIZE);
dee41079	167	if (phys == -1) {
6daaca52	168	dev_dbg(dev, "pgoff_to_phys(%#lx) failed\n", pgoff);
dee41079 DW	169	return VM_FAULT_SIGBUS;
	170	}
	171
6ec228b6	172	pfn = phys_to_pfn_t(phys, PFN_DEV\|PFN_MAP);
dee41079	173
6ec228b6	174	dax_set_mapping(vmf, pfn, fault_size);
0e7325f0	175
6ec228b6	176	return vmf_insert_pfn_pmd(vmf, pfn, vmf->flags & FAULT_FLAG_WRITE);
dee41079 DW	177	}
dee41079 DW	178
9557feee	179	#ifdef CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD
226ab561	180	static vm_fault_t __dev_dax_pud_fault(struct dev_dax *dev_dax,
6ec228b6	181	struct vm_fault *vmf)
9557feee DJ	182	{
9557feee DJ	183	unsigned long pud_addr = vmf->address & PUD_MASK;
5f0694b3	184	struct device *dev = &dev_dax->dev;
9557feee DJ	185	phys_addr_t phys;
9557feee DJ	186	pgoff_t pgoff;
6ec228b6	187	pfn_t pfn;
70b085b0 DJ	188	unsigned int fault_size = PUD_SIZE;
70b085b0 DJ	189
9557feee	190
5f0694b3	191	if (check_vma(dev_dax, vmf->vma, __func__))
9557feee DJ	192	return VM_FAULT_SIGBUS;
9557feee DJ	193
33cf94d7	194	if (dev_dax->align > PUD_SIZE) {
6daaca52	195	dev_dbg(dev, "alignment (%#x) > fault size (%#x)\n",
33cf94d7	196	dev_dax->align, fault_size);
9557feee DJ	197	return VM_FAULT_SIGBUS;
	198	}
	199
33cf94d7	200	if (fault_size < dev_dax->align)
70b085b0	201	return VM_FAULT_SIGBUS;
33cf94d7	202	else if (fault_size > dev_dax->align)
70b085b0 DJ	203	return VM_FAULT_FALLBACK;
	204
	205	/* if we are outside of the VMA */
	206	if (pud_addr < vmf->vma->vm_start \|\|
	207	(pud_addr + PUD_SIZE) > vmf->vma->vm_end)
	208	return VM_FAULT_SIGBUS;
	209
9557feee	210	pgoff = linear_page_index(vmf->vma, pud_addr);
73616367	211	phys = dax_pgoff_to_phys(dev_dax, pgoff, PUD_SIZE);
9557feee	212	if (phys == -1) {
6daaca52	213	dev_dbg(dev, "pgoff_to_phys(%#lx) failed\n", pgoff);
9557feee DJ	214	return VM_FAULT_SIGBUS;
	215	}
	216
6ec228b6	217	pfn = phys_to_pfn_t(phys, PFN_DEV\|PFN_MAP);
9557feee	218
6ec228b6	219	dax_set_mapping(vmf, pfn, fault_size);
0e7325f0	220
6ec228b6	221	return vmf_insert_pfn_pud(vmf, pfn, vmf->flags & FAULT_FLAG_WRITE);
9557feee DJ	222	}
9557feee DJ	223	#else
226ab561	224	static vm_fault_t __dev_dax_pud_fault(struct dev_dax *dev_dax,
6ec228b6	225	struct vm_fault *vmf)
9557feee DJ	226	{
	227	return VM_FAULT_FALLBACK;
	228	}
	229	#endif /* !CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD */
	230
226ab561	231	static vm_fault_t dev_dax_huge_fault(struct vm_fault *vmf,
c791ace1	232	enum page_entry_size pe_size)
dee41079	233	{
f4200391	234	struct file *filp = vmf->vma->vm_file;
36bdac1e SJ	235	vm_fault_t rc = VM_FAULT_SIGBUS;
36bdac1e SJ	236	int id;
5f0694b3	237	struct dev_dax *dev_dax = filp->private_data;
dee41079	238
6daaca52 DW	239	dev_dbg(&dev_dax->dev, "%s: %s (%#lx - %#lx) size = %d\n", current->comm,
6daaca52 DW	240	(vmf->flags & FAULT_FLAG_WRITE) ? "write" : "read",
76202620	241	vmf->vma->vm_start, vmf->vma->vm_end, pe_size);
dee41079	242
7b6be844	243	id = dax_read_lock();
c791ace1 DJ	244	switch (pe_size) {
c791ace1 DJ	245	case PE_SIZE_PTE:
6ec228b6	246	rc = __dev_dax_pte_fault(dev_dax, vmf);
a2d58167	247	break;
c791ace1	248	case PE_SIZE_PMD:
6ec228b6	249	rc = __dev_dax_pmd_fault(dev_dax, vmf);
9557feee	250	break;
c791ace1	251	case PE_SIZE_PUD:
6ec228b6	252	rc = __dev_dax_pud_fault(dev_dax, vmf);
a2d58167 DJ	253	break;
a2d58167 DJ	254	default:
54eafcc9	255	rc = VM_FAULT_SIGBUS;
a2d58167	256	}
2232c638	257
7b6be844	258	dax_read_unlock(id);
dee41079 DW	259
	260	return rc;
	261	}
	262
226ab561	263	static vm_fault_t dev_dax_fault(struct vm_fault *vmf)
c791ace1	264	{
5f0694b3	265	return dev_dax_huge_fault(vmf, PE_SIZE_PTE);
c791ace1 DJ	266	}
c791ace1 DJ	267
dd3b614f	268	static int dev_dax_may_split(struct vm_area_struct *vma, unsigned long addr)
9702cffd DW	269	{
	270	struct file *filp = vma->vm_file;
	271	struct dev_dax *dev_dax = filp->private_data;
9702cffd	272
33cf94d7	273	if (!IS_ALIGNED(addr, dev_dax->align))
9702cffd DW	274	return -EINVAL;
	275	return 0;
	276	}
	277
c1d53b92 DW	278	static unsigned long dev_dax_pagesize(struct vm_area_struct *vma)
	279	{
	280	struct file *filp = vma->vm_file;
	281	struct dev_dax *dev_dax = filp->private_data;
c1d53b92	282
33cf94d7	283	return dev_dax->align;
c1d53b92 DW	284	}
c1d53b92 DW	285
5f0694b3 DW	286	static const struct vm_operations_struct dax_vm_ops = {
	287	.fault = dev_dax_fault,
	288	.huge_fault = dev_dax_huge_fault,
dd3b614f	289	.may_split = dev_dax_may_split,
c1d53b92	290	.pagesize = dev_dax_pagesize,
dee41079 DW	291	};
dee41079 DW	292
af69f51e	293	static int dax_mmap(struct file filp, struct vm_area_struct vma)
dee41079	294	{
5f0694b3	295	struct dev_dax *dev_dax = filp->private_data;
7b6be844	296	int rc, id;
dee41079	297
6daaca52	298	dev_dbg(&dev_dax->dev, "trace\n");
dee41079	299
7b6be844 DW	300	/*
	301	* We lock to check dax_dev liveness and will re-check at
	302	* fault time.
	303	*/
	304	id = dax_read_lock();
5f0694b3	305	rc = check_vma(dev_dax, vma, __func__);
7b6be844	306	dax_read_unlock(id);
dee41079 DW	307	if (rc)
	308	return rc;
	309
5f0694b3	310	vma->vm_ops = &dax_vm_ops;
e1fb4a08	311	vma->vm_flags \|= VM_HUGEPAGE;
dee41079	312	return 0;
043a9255 DW	313	}
	314
	315	/* return an unmapped area aligned to the dax region specified alignment */
af69f51e	316	static unsigned long dax_get_unmapped_area(struct file *filp,
043a9255 DW	317	unsigned long addr, unsigned long len, unsigned long pgoff,
	318	unsigned long flags)
	319	{
	320	unsigned long off, off_end, off_align, len_align, addr_align, align;
5f0694b3	321	struct dev_dax *dev_dax = filp ? filp->private_data : NULL;
043a9255	322
5f0694b3	323	if (!dev_dax \|\| addr)
043a9255 DW	324	goto out;
043a9255 DW	325
33cf94d7	326	align = dev_dax->align;
043a9255 DW	327	off = pgoff << PAGE_SHIFT;
	328	off_end = off + len;
	329	off_align = round_up(off, align);
	330
	331	if ((off_end <= off_align) \|\| ((off_end - off_align) < align))
	332	goto out;
	333
	334	len_align = len + align;
	335	if ((off + len_align) < off)
	336	goto out;
dee41079	337
043a9255 DW	338	addr_align = current->mm->get_unmapped_area(filp, addr, len_align,
	339	pgoff, flags);
	340	if (!IS_ERR_VALUE(addr_align)) {
	341	addr_align += (off - addr_align) & (align - 1);
	342	return addr_align;
	343	}
	344	out:
	345	return current->mm->get_unmapped_area(filp, addr, len, pgoff, flags);
	346	}
	347
41c9b1be	348	static const struct address_space_operations dev_dax_aops = {
46de8b97	349	.dirty_folio = noop_dirty_folio,
41c9b1be DJ	350	};
41c9b1be DJ	351
af69f51e	352	static int dax_open(struct inode inode, struct file filp)
043a9255	353	{
7b6be844 DW	354	struct dax_device *dax_dev = inode_dax(inode);
	355	struct inode *__dax_inode = dax_inode(dax_dev);
	356	struct dev_dax *dev_dax = dax_get_private(dax_dev);
043a9255	357
6daaca52	358	dev_dbg(&dev_dax->dev, "trace\n");
7b6be844 DW	359	inode->i_mapping = __dax_inode->i_mapping;
7b6be844 DW	360	inode->i_mapping->host = __dax_inode;
41c9b1be	361	inode->i_mapping->a_ops = &dev_dax_aops;
3bc52c45	362	filp->f_mapping = inode->i_mapping;
5660e13d	363	filp->f_wb_err = filemap_sample_wb_err(filp->f_mapping);
735e4ae5	364	filp->f_sb_err = file_sample_sb_err(filp);
5f0694b3	365	filp->private_data = dev_dax;
ebd84d72	366	inode->i_flags = S_DAX;
043a9255	367
043a9255 DW	368	return 0;
043a9255 DW	369	}
dee41079	370
af69f51e	371	static int dax_release(struct inode inode, struct file filp)
043a9255	372	{
5f0694b3	373	struct dev_dax *dev_dax = filp->private_data;
043a9255	374
6daaca52	375	dev_dbg(&dev_dax->dev, "trace\n");
043a9255	376	return 0;
dee41079 DW	377	}
dee41079 DW	378
ab68f262 DW	379	static const struct file_operations dax_fops = {
	380	.llseek = noop_llseek,
	381	.owner = THIS_MODULE,
af69f51e DW	382	.open = dax_open,
	383	.release = dax_release,
	384	.get_unmapped_area = dax_get_unmapped_area,
	385	.mmap = dax_mmap,
ef842302	386	.mmap_supported_flags = MAP_SYNC,
ab68f262 DW	387	};
ab68f262 DW	388
9567da0b	389	static void dev_dax_cdev_del(void *cdev)
043a9255	390	{
9567da0b DW	391	cdev_del(cdev);
9567da0b DW	392	}
043a9255	393
9567da0b DW	394	static void dev_dax_kill(void *dev_dax)
	395	{
	396	kill_dev_dax(dev_dax);
ebd84d72 DW	397	}
ebd84d72 DW	398
f11cf813	399	int dev_dax_probe(struct dev_dax *dev_dax)
043a9255	400	{
9567da0b	401	struct dax_device *dax_dev = dev_dax->dax_dev;
f11cf813	402	struct device *dev = &dev_dax->dev;
f5516ec5	403	struct dev_pagemap *pgmap;
7b6be844	404	struct inode *inode;
ba09c01d	405	struct cdev *cdev;
89ec9f2c	406	void *addr;
60e93dc0	407	int rc, i;
89ec9f2c	408
fc65c4eb JM	409	if (static_dev_dax(dev_dax)) {
	410	if (dev_dax->nr_range > 1) {
	411	dev_warn(dev,
	412	"static pgmap / multi-range device conflict\n");
	413	return -EINVAL;
	414	}
60e93dc0	415
fc65c4eb JM	416	pgmap = dev_dax->pgmap;
	417	} else {
	418	if (dev_dax->pgmap) {
	419	dev_warn(dev,
	420	"dynamic-dax with pre-populated page map\n");
	421	return -EINVAL;
	422	}
60e93dc0	423
09b80137 JM	424	pgmap = devm_kzalloc(dev,
	425	struct_size(pgmap, ranges, dev_dax->nr_range - 1),
	426	GFP_KERNEL);
f5516ec5 DW	427	if (!pgmap)
f5516ec5 DW	428	return -ENOMEM;
fc65c4eb	429
60e93dc0	430	pgmap->nr_range = dev_dax->nr_range;
fc65c4eb JM	431	dev_dax->pgmap = pgmap;
	432
	433	for (i = 0; i < dev_dax->nr_range; i++) {
	434	struct range *range = &dev_dax->ranges[i].range;
	435	pgmap->ranges[i] = *range;
	436	}
60e93dc0 DW	437	}
	438
	439	for (i = 0; i < dev_dax->nr_range; i++) {
	440	struct range *range = &dev_dax->ranges[i].range;
	441
	442	if (!devm_request_mem_region(dev, range->start,
	443	range_len(range), dev_name(dev))) {
	444	dev_warn(dev, "mapping%d: %#llx-%#llx could not reserve range\n",
	445	i, range->start, range->end);
	446	return -EBUSY;
	447	}
f5516ec5	448	}
60e93dc0	449
f5516ec5	450	pgmap->type = MEMORY_DEVICE_GENERIC;
14606001 JM	451	if (dev_dax->align > PAGE_SIZE)
	452	pgmap->vmemmap_shift =
	453	order_base_2(dev_dax->align >> PAGE_SHIFT);
f5516ec5	454	addr = devm_memremap_pages(dev, pgmap);
50f44ee7	455	if (IS_ERR(addr))
89ec9f2c	456	return PTR_ERR(addr);
89ec9f2c	457
7b6be844 DW	458	inode = dax_inode(dax_dev);
7b6be844 DW	459	cdev = inode->i_cdev;
ba09c01d	460	cdev_init(cdev, &dax_fops);
83762cb5	461	cdev->owner = dev->driver->owner;
9567da0b DW	462	cdev_set_parent(cdev, &dev->kobj);
9567da0b DW	463	rc = cdev_add(cdev, dev->devt, 1);
d76911ee	464	if (rc)
9567da0b	465	return rc;
d76911ee	466
9567da0b DW	467	rc = devm_add_action_or_reset(dev, dev_dax_cdev_del, cdev);
	468	if (rc)
	469	return rc;
043a9255	470
9567da0b DW	471	run_dax(dax_dev);
	472	return devm_add_action_or_reset(dev, dev_dax_kill, dev_dax);
	473	}
730926c3	474	EXPORT_SYMBOL_GPL(dev_dax_probe);
043a9255	475
d200781e	476	static struct dax_device_driver device_dax_driver = {
f11cf813	477	.probe = dev_dax_probe,
c80b5320	478	/* all probe actions are unwound by devm, so .remove isn't necessary */
d200781e	479	.match_always = 1,
9567da0b	480	};
043a9255	481
ab68f262 DW	482	static int __init dax_init(void)
ab68f262 DW	483	{
9567da0b	484	return dax_driver_register(&device_dax_driver);
ab68f262 DW	485	}
	486
	487	static void __exit dax_exit(void)
	488	{
d200781e	489	dax_driver_unregister(&device_dax_driver);
ab68f262 DW	490	}
	491
	492	MODULE_AUTHOR("Intel Corporation");
	493	MODULE_LICENSE("GPL v2");
9567da0b	494	module_init(dax_init);
ab68f262	495	module_exit(dax_exit);
9567da0b	496	MODULE_ALIAS_DAX_DEVICE(0);