[linux-2.6-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_DOWN(vmf->address, fault_size));

	for (i = 0; i < nr_pages; i++) {
		struct folio *folio = pfn_folio(pfn_t_to_pfn(pfn) + i);

		if (folio->mapping)
			continue;

		folio->mapping = filp->f_mapping;
		folio->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, 0);

	dax_set_mapping(vmf, pfn, fault_size);

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

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, 0);

	dax_set_mapping(vmf, pfn, fault_size);

	return vmf_insert_folio_pmd(vmf, page_folio(pfn_t_to_page(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, 0);

	dax_set_mapping(vmf, pfn, fault_size);

	return vmf_insert_folio_pud(vmf, page_folio(pfn_t_to_page(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, unsigned int order)
{
	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: op=%s addr=%#lx order=%d\n", current->comm,
		(vmf->flags & FAULT_FLAG_WRITE) ? "write" : "read",
		vmf->address & ~((1UL << (order + PAGE_SHIFT)) - 1), order);

	id = dax_read_lock();
	if (order == 0)
		rc = __dev_dax_pte_fault(dev_dax, vmf);
	else if (order == PMD_ORDER)
		rc = __dev_dax_pmd_fault(dev_dax, vmf);
	else if (order == PUD_ORDER)
		rc = __dev_dax_pud_fault(dev_dax, vmf);
	else
		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, 0);
}

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;
	vm_flags_set(vma, 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 = mm_get_unmapped_area(current->mm, filp, addr, len_align,
					  pgoff, flags);
	if (!IS_ERR_VALUE(addr_align)) {
		addr_align += (off - addr_align) & (align - 1);
		return addr_align;
	}
 out:
	return mm_get_unmapped_area(current->mm, 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,
	.fop_flags = FOP_MMAP_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);
}

static 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);
}

static struct dax_device_driver device_dax_driver = {
	.probe = dev_dax_probe,
	.type = DAXDRV_DEVICE_TYPE,
};

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_DESCRIPTION("Device DAX: direct access device driver");
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	88	pgoff = linear_page_index(vmf->vma,
7fcbd978	89	ALIGN_DOWN(vmf->address, fault_size));
a0fb038e JM	90
a0fb038e JM	91	for (i = 0; i < nr_pages; i++) {
995abaaa	92	struct folio *folio = pfn_folio(pfn_t_to_pfn(pfn) + i);
a0fb038e	93
995abaaa	94	if (folio->mapping)
a0fb038e JM	95	continue;
a0fb038e JM	96
995abaaa MWO	97	folio->mapping = filp->f_mapping;
995abaaa MWO	98	folio->index = pgoff + i;
a0fb038e JM	99	}
	100	}
	101
226ab561	102	static vm_fault_t __dev_dax_pte_fault(struct dev_dax *dev_dax,
6ec228b6	103	struct vm_fault *vmf)
dee41079	104	{
5f0694b3	105	struct device *dev = &dev_dax->dev;
dee41079	106	phys_addr_t phys;
6ec228b6	107	pfn_t pfn;
0134ed4f	108	unsigned int fault_size = PAGE_SIZE;
dee41079	109
5f0694b3	110	if (check_vma(dev_dax, vmf->vma, __func__))
dee41079 DW	111	return VM_FAULT_SIGBUS;
dee41079 DW	112
33cf94d7	113	if (dev_dax->align > PAGE_SIZE) {
6daaca52	114	dev_dbg(dev, "alignment (%#x) > fault size (%#x)\n",
33cf94d7	115	dev_dax->align, fault_size);
dee41079 DW	116	return VM_FAULT_SIGBUS;
	117	}
	118
33cf94d7	119	if (fault_size != dev_dax->align)
0134ed4f DJ	120	return VM_FAULT_SIGBUS;
0134ed4f DJ	121
73616367	122	phys = dax_pgoff_to_phys(dev_dax, vmf->pgoff, PAGE_SIZE);
dee41079	123	if (phys == -1) {
6daaca52	124	dev_dbg(dev, "pgoff_to_phys(%#lx) failed\n", vmf->pgoff);
dee41079 DW	125	return VM_FAULT_SIGBUS;
	126	}
	127
aed877c2	128	pfn = phys_to_pfn_t(phys, 0);
dee41079	129
6ec228b6	130	dax_set_mapping(vmf, pfn, fault_size);
0e7325f0	131
aed877c2 AP	132	return vmf_insert_page_mkwrite(vmf, pfn_t_to_page(pfn),
aed877c2 AP	133	vmf->flags & FAULT_FLAG_WRITE);
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
aed877c2	172	pfn = phys_to_pfn_t(phys, 0);
dee41079	173
6ec228b6	174	dax_set_mapping(vmf, pfn, fault_size);
0e7325f0	175
aed877c2 AP	176	return vmf_insert_folio_pmd(vmf, page_folio(pfn_t_to_page(pfn)),
aed877c2 AP	177	vmf->flags & FAULT_FLAG_WRITE);
dee41079 DW	178	}
dee41079 DW	179
9557feee	180	#ifdef CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD
226ab561	181	static vm_fault_t __dev_dax_pud_fault(struct dev_dax *dev_dax,
6ec228b6	182	struct vm_fault *vmf)
9557feee DJ	183	{
9557feee DJ	184	unsigned long pud_addr = vmf->address & PUD_MASK;
5f0694b3	185	struct device *dev = &dev_dax->dev;
9557feee DJ	186	phys_addr_t phys;
9557feee DJ	187	pgoff_t pgoff;
6ec228b6	188	pfn_t pfn;
70b085b0 DJ	189	unsigned int fault_size = PUD_SIZE;
70b085b0 DJ	190
9557feee	191
5f0694b3	192	if (check_vma(dev_dax, vmf->vma, __func__))
9557feee DJ	193	return VM_FAULT_SIGBUS;
9557feee DJ	194
33cf94d7	195	if (dev_dax->align > PUD_SIZE) {
6daaca52	196	dev_dbg(dev, "alignment (%#x) > fault size (%#x)\n",
33cf94d7	197	dev_dax->align, fault_size);
9557feee DJ	198	return VM_FAULT_SIGBUS;
	199	}
	200
33cf94d7	201	if (fault_size < dev_dax->align)
70b085b0	202	return VM_FAULT_SIGBUS;
33cf94d7	203	else if (fault_size > dev_dax->align)
70b085b0 DJ	204	return VM_FAULT_FALLBACK;
	205
	206	/* if we are outside of the VMA */
	207	if (pud_addr < vmf->vma->vm_start \|\|
	208	(pud_addr + PUD_SIZE) > vmf->vma->vm_end)
	209	return VM_FAULT_SIGBUS;
	210
9557feee	211	pgoff = linear_page_index(vmf->vma, pud_addr);
73616367	212	phys = dax_pgoff_to_phys(dev_dax, pgoff, PUD_SIZE);
9557feee	213	if (phys == -1) {
6daaca52	214	dev_dbg(dev, "pgoff_to_phys(%#lx) failed\n", pgoff);
9557feee DJ	215	return VM_FAULT_SIGBUS;
	216	}
	217
aed877c2	218	pfn = phys_to_pfn_t(phys, 0);
9557feee	219
6ec228b6	220	dax_set_mapping(vmf, pfn, fault_size);
0e7325f0	221
aed877c2 AP	222	return vmf_insert_folio_pud(vmf, page_folio(pfn_t_to_page(pfn)),
aed877c2 AP	223	vmf->flags & FAULT_FLAG_WRITE);
9557feee DJ	224	}
9557feee DJ	225	#else
226ab561	226	static vm_fault_t __dev_dax_pud_fault(struct dev_dax *dev_dax,
6ec228b6	227	struct vm_fault *vmf)
9557feee DJ	228	{
	229	return VM_FAULT_FALLBACK;
	230	}
	231	#endif /* !CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD */
	232
1d024e7a	233	static vm_fault_t dev_dax_huge_fault(struct vm_fault *vmf, unsigned int order)
dee41079	234	{
f4200391	235	struct file *filp = vmf->vma->vm_file;
36bdac1e SJ	236	vm_fault_t rc = VM_FAULT_SIGBUS;
36bdac1e SJ	237	int id;
5f0694b3	238	struct dev_dax *dev_dax = filp->private_data;
dee41079	239
5b198b47 PX	240	dev_dbg(&dev_dax->dev, "%s: op=%s addr=%#lx order=%d\n", current->comm,
	241	(vmf->flags & FAULT_FLAG_WRITE) ? "write" : "read",
	242	vmf->address & ~((1UL << (order + PAGE_SHIFT)) - 1), order);
dee41079	243
7b6be844	244	id = dax_read_lock();
1d024e7a	245	if (order == 0)
6ec228b6	246	rc = __dev_dax_pte_fault(dev_dax, vmf);
1d024e7a	247	else if (order == PMD_ORDER)
6ec228b6	248	rc = __dev_dax_pmd_fault(dev_dax, vmf);
1d024e7a	249	else if (order == PUD_ORDER)
6ec228b6	250	rc = __dev_dax_pud_fault(dev_dax, vmf);
1d024e7a	251	else
54eafcc9	252	rc = VM_FAULT_SIGBUS;
2232c638	253
7b6be844	254	dax_read_unlock(id);
dee41079 DW	255
	256	return rc;
	257	}
	258
226ab561	259	static vm_fault_t dev_dax_fault(struct vm_fault *vmf)
c791ace1	260	{
1d024e7a	261	return dev_dax_huge_fault(vmf, 0);
c791ace1 DJ	262	}
c791ace1 DJ	263
dd3b614f	264	static int dev_dax_may_split(struct vm_area_struct *vma, unsigned long addr)
9702cffd DW	265	{
	266	struct file *filp = vma->vm_file;
	267	struct dev_dax *dev_dax = filp->private_data;
9702cffd	268
33cf94d7	269	if (!IS_ALIGNED(addr, dev_dax->align))
9702cffd DW	270	return -EINVAL;
	271	return 0;
	272	}
	273
c1d53b92 DW	274	static unsigned long dev_dax_pagesize(struct vm_area_struct *vma)
	275	{
	276	struct file *filp = vma->vm_file;
	277	struct dev_dax *dev_dax = filp->private_data;
c1d53b92	278
33cf94d7	279	return dev_dax->align;
c1d53b92 DW	280	}
c1d53b92 DW	281
5f0694b3 DW	282	static const struct vm_operations_struct dax_vm_ops = {
	283	.fault = dev_dax_fault,
	284	.huge_fault = dev_dax_huge_fault,
dd3b614f	285	.may_split = dev_dax_may_split,
c1d53b92	286	.pagesize = dev_dax_pagesize,
dee41079 DW	287	};
dee41079 DW	288
af69f51e	289	static int dax_mmap(struct file filp, struct vm_area_struct vma)
dee41079	290	{
5f0694b3	291	struct dev_dax *dev_dax = filp->private_data;
7b6be844	292	int rc, id;
dee41079	293
6daaca52	294	dev_dbg(&dev_dax->dev, "trace\n");
dee41079	295
7b6be844 DW	296	/*
	297	* We lock to check dax_dev liveness and will re-check at
	298	* fault time.
	299	*/
	300	id = dax_read_lock();
5f0694b3	301	rc = check_vma(dev_dax, vma, __func__);
7b6be844	302	dax_read_unlock(id);
dee41079 DW	303	if (rc)
	304	return rc;
	305
5f0694b3	306	vma->vm_ops = &dax_vm_ops;
1c71222e	307	vm_flags_set(vma, VM_HUGEPAGE);
dee41079	308	return 0;
043a9255 DW	309	}
	310
	311	/* return an unmapped area aligned to the dax region specified alignment */
af69f51e	312	static unsigned long dax_get_unmapped_area(struct file *filp,
043a9255 DW	313	unsigned long addr, unsigned long len, unsigned long pgoff,
	314	unsigned long flags)
	315	{
	316	unsigned long off, off_end, off_align, len_align, addr_align, align;
5f0694b3	317	struct dev_dax *dev_dax = filp ? filp->private_data : NULL;
043a9255	318
5f0694b3	319	if (!dev_dax \|\| addr)
043a9255 DW	320	goto out;
043a9255 DW	321
33cf94d7	322	align = dev_dax->align;
043a9255 DW	323	off = pgoff << PAGE_SHIFT;
	324	off_end = off + len;
	325	off_align = round_up(off, align);
	326
	327	if ((off_end <= off_align) \|\| ((off_end - off_align) < align))
	328	goto out;
	329
	330	len_align = len + align;
	331	if ((off + len_align) < off)
	332	goto out;
dee41079	333
529ce23a RE	334	addr_align = mm_get_unmapped_area(current->mm, filp, addr, len_align,
529ce23a RE	335	pgoff, flags);
043a9255 DW	336	if (!IS_ERR_VALUE(addr_align)) {
	337	addr_align += (off - addr_align) & (align - 1);
	338	return addr_align;
	339	}
	340	out:
529ce23a	341	return mm_get_unmapped_area(current->mm, filp, addr, len, pgoff, flags);
043a9255 DW	342	}
043a9255 DW	343
41c9b1be	344	static const struct address_space_operations dev_dax_aops = {
46de8b97	345	.dirty_folio = noop_dirty_folio,
41c9b1be DJ	346	};
41c9b1be DJ	347
af69f51e	348	static int dax_open(struct inode inode, struct file filp)
043a9255	349	{
7b6be844 DW	350	struct dax_device *dax_dev = inode_dax(inode);
	351	struct inode *__dax_inode = dax_inode(dax_dev);
	352	struct dev_dax *dev_dax = dax_get_private(dax_dev);
043a9255	353
6daaca52	354	dev_dbg(&dev_dax->dev, "trace\n");
7b6be844 DW	355	inode->i_mapping = __dax_inode->i_mapping;
7b6be844 DW	356	inode->i_mapping->host = __dax_inode;
41c9b1be	357	inode->i_mapping->a_ops = &dev_dax_aops;
3bc52c45	358	filp->f_mapping = inode->i_mapping;
5660e13d	359	filp->f_wb_err = filemap_sample_wb_err(filp->f_mapping);
735e4ae5	360	filp->f_sb_err = file_sample_sb_err(filp);
5f0694b3	361	filp->private_data = dev_dax;
ebd84d72	362	inode->i_flags = S_DAX;
043a9255	363
043a9255 DW	364	return 0;
043a9255 DW	365	}
dee41079	366
af69f51e	367	static int dax_release(struct inode inode, struct file filp)
043a9255	368	{
5f0694b3	369	struct dev_dax *dev_dax = filp->private_data;
043a9255	370
6daaca52	371	dev_dbg(&dev_dax->dev, "trace\n");
043a9255	372	return 0;
dee41079 DW	373	}
dee41079 DW	374
ab68f262 DW	375	static const struct file_operations dax_fops = {
	376	.llseek = noop_llseek,
	377	.owner = THIS_MODULE,
af69f51e DW	378	.open = dax_open,
	379	.release = dax_release,
	380	.get_unmapped_area = dax_get_unmapped_area,
	381	.mmap = dax_mmap,
210a03c9	382	.fop_flags = FOP_MMAP_SYNC,
ab68f262 DW	383	};
ab68f262 DW	384
9567da0b	385	static void dev_dax_cdev_del(void *cdev)
043a9255	386	{
9567da0b DW	387	cdev_del(cdev);
9567da0b DW	388	}
043a9255	389
9567da0b DW	390	static void dev_dax_kill(void *dev_dax)
	391	{
	392	kill_dev_dax(dev_dax);
ebd84d72 DW	393	}
ebd84d72 DW	394
2d515352	395	static int dev_dax_probe(struct dev_dax *dev_dax)
043a9255	396	{
9567da0b	397	struct dax_device *dax_dev = dev_dax->dax_dev;
f11cf813	398	struct device *dev = &dev_dax->dev;
f5516ec5	399	struct dev_pagemap *pgmap;
7b6be844	400	struct inode *inode;
ba09c01d	401	struct cdev *cdev;
89ec9f2c	402	void *addr;
60e93dc0	403	int rc, i;
89ec9f2c	404
fc65c4eb JM	405	if (static_dev_dax(dev_dax)) {
	406	if (dev_dax->nr_range > 1) {
	407	dev_warn(dev,
	408	"static pgmap / multi-range device conflict\n");
	409	return -EINVAL;
	410	}
60e93dc0	411
fc65c4eb JM	412	pgmap = dev_dax->pgmap;
	413	} else {
	414	if (dev_dax->pgmap) {
	415	dev_warn(dev,
	416	"dynamic-dax with pre-populated page map\n");
	417	return -EINVAL;
	418	}
60e93dc0	419
09b80137 JM	420	pgmap = devm_kzalloc(dev,
	421	struct_size(pgmap, ranges, dev_dax->nr_range - 1),
	422	GFP_KERNEL);
f5516ec5 DW	423	if (!pgmap)
f5516ec5 DW	424	return -ENOMEM;
fc65c4eb	425
60e93dc0	426	pgmap->nr_range = dev_dax->nr_range;
fc65c4eb JM	427	dev_dax->pgmap = pgmap;
	428
	429	for (i = 0; i < dev_dax->nr_range; i++) {
	430	struct range *range = &dev_dax->ranges[i].range;
	431	pgmap->ranges[i] = *range;
	432	}
60e93dc0 DW	433	}
	434
	435	for (i = 0; i < dev_dax->nr_range; i++) {
	436	struct range *range = &dev_dax->ranges[i].range;
	437
	438	if (!devm_request_mem_region(dev, range->start,
	439	range_len(range), dev_name(dev))) {
	440	dev_warn(dev, "mapping%d: %#llx-%#llx could not reserve range\n",
	441	i, range->start, range->end);
	442	return -EBUSY;
	443	}
f5516ec5	444	}
60e93dc0	445
f5516ec5	446	pgmap->type = MEMORY_DEVICE_GENERIC;
14606001 JM	447	if (dev_dax->align > PAGE_SIZE)
	448	pgmap->vmemmap_shift =
	449	order_base_2(dev_dax->align >> PAGE_SHIFT);
f5516ec5	450	addr = devm_memremap_pages(dev, pgmap);
50f44ee7	451	if (IS_ERR(addr))
89ec9f2c	452	return PTR_ERR(addr);
89ec9f2c	453
7b6be844 DW	454	inode = dax_inode(dax_dev);
7b6be844 DW	455	cdev = inode->i_cdev;
ba09c01d	456	cdev_init(cdev, &dax_fops);
83762cb5	457	cdev->owner = dev->driver->owner;
9567da0b DW	458	cdev_set_parent(cdev, &dev->kobj);
9567da0b DW	459	rc = cdev_add(cdev, dev->devt, 1);
d76911ee	460	if (rc)
9567da0b	461	return rc;
d76911ee	462
9567da0b DW	463	rc = devm_add_action_or_reset(dev, dev_dax_cdev_del, cdev);
	464	if (rc)
	465	return rc;
043a9255	466
9567da0b DW	467	run_dax(dax_dev);
	468	return devm_add_action_or_reset(dev, dev_dax_kill, dev_dax);
	469	}
043a9255	470
d200781e	471	static struct dax_device_driver device_dax_driver = {
f11cf813	472	.probe = dev_dax_probe,
e9ee9fe3	473	.type = DAXDRV_DEVICE_TYPE,
9567da0b	474	};
043a9255	475
ab68f262 DW	476	static int __init dax_init(void)
ab68f262 DW	477	{
9567da0b	478	return dax_driver_register(&device_dax_driver);
ab68f262 DW	479	}
	480
	481	static void __exit dax_exit(void)
	482	{
d200781e	483	dax_driver_unregister(&device_dax_driver);
ab68f262 DW	484	}
	485
	486	MODULE_AUTHOR("Intel Corporation");
1d5198dd	487	MODULE_DESCRIPTION("Device DAX: direct access device driver");
ab68f262	488	MODULE_LICENSE("GPL v2");
9567da0b	489	module_init(dax_init);
ab68f262	490	module_exit(dax_exit);
9567da0b	491	MODULE_ALIAS_DAX_DEVICE(0);