[linux-block.git] / fs / erofs / data.c

// SPDX-License-Identifier: GPL-2.0-only
/*
 * Copyright (C) 2017-2018 HUAWEI, Inc.
 *             https://www.huawei.com/
 * Copyright (C) 2021, Alibaba Cloud
 */
#include "internal.h"
#include <linux/prefetch.h>
#include <linux/sched/mm.h>
#include <linux/dax.h>
#include <trace/events/erofs.h>

void erofs_unmap_metabuf(struct erofs_buf *buf)
{
	if (buf->kmap_type == EROFS_KMAP)
		kunmap_local(buf->base);
	buf->base = NULL;
	buf->kmap_type = EROFS_NO_KMAP;
}

void erofs_put_metabuf(struct erofs_buf *buf)
{
	if (!buf->page)
		return;
	erofs_unmap_metabuf(buf);
	put_page(buf->page);
	buf->page = NULL;
}

/*
 * Derive the block size from inode->i_blkbits to make compatible with
 * anonymous inode in fscache mode.
 */
void *erofs_bread(struct erofs_buf *buf, erofs_blk_t blkaddr,
		  enum erofs_kmap_type type)
{
	struct inode *inode = buf->inode;
	erofs_off_t offset = (erofs_off_t)blkaddr << inode->i_blkbits;
	pgoff_t index = offset >> PAGE_SHIFT;
	struct page *page = buf->page;
	struct folio *folio;
	unsigned int nofs_flag;

	if (!page || page->index != index) {
		erofs_put_metabuf(buf);

		nofs_flag = memalloc_nofs_save();
		folio = read_cache_folio(inode->i_mapping, index, NULL, NULL);
		memalloc_nofs_restore(nofs_flag);
		if (IS_ERR(folio))
			return folio;

		/* should already be PageUptodate, no need to lock page */
		page = folio_file_page(folio, index);
		buf->page = page;
	}
	if (buf->kmap_type == EROFS_NO_KMAP) {
		if (type == EROFS_KMAP)
			buf->base = kmap_local_page(page);
		buf->kmap_type = type;
	} else if (buf->kmap_type != type) {
		DBG_BUGON(1);
		return ERR_PTR(-EFAULT);
	}
	if (type == EROFS_NO_KMAP)
		return NULL;
	return buf->base + (offset & ~PAGE_MASK);
}

void erofs_init_metabuf(struct erofs_buf *buf, struct super_block *sb)
{
	if (erofs_is_fscache_mode(sb))
		buf->inode = EROFS_SB(sb)->s_fscache->inode;
	else
		buf->inode = sb->s_bdev->bd_inode;
}

void *erofs_read_metabuf(struct erofs_buf *buf, struct super_block *sb,
			 erofs_blk_t blkaddr, enum erofs_kmap_type type)
{
	erofs_init_metabuf(buf, sb);
	return erofs_bread(buf, blkaddr, type);
}

static int erofs_map_blocks_flatmode(struct inode *inode,
				     struct erofs_map_blocks *map)
{
	erofs_blk_t nblocks, lastblk;
	u64 offset = map->m_la;
	struct erofs_inode *vi = EROFS_I(inode);
	struct super_block *sb = inode->i_sb;
	bool tailendpacking = (vi->datalayout == EROFS_INODE_FLAT_INLINE);

	nblocks = erofs_iblks(inode);
	lastblk = nblocks - tailendpacking;

	/* there is no hole in flatmode */
	map->m_flags = EROFS_MAP_MAPPED;
	if (offset < erofs_pos(sb, lastblk)) {
		map->m_pa = erofs_pos(sb, vi->raw_blkaddr) + map->m_la;
		map->m_plen = erofs_pos(sb, lastblk) - offset;
	} else if (tailendpacking) {
		map->m_pa = erofs_iloc(inode) + vi->inode_isize +
			vi->xattr_isize + erofs_blkoff(sb, offset);
		map->m_plen = inode->i_size - offset;

		/* inline data should be located in the same meta block */
		if (erofs_blkoff(sb, map->m_pa) + map->m_plen > sb->s_blocksize) {
			erofs_err(sb, "inline data cross block boundary @ nid %llu",
				  vi->nid);
			DBG_BUGON(1);
			return -EFSCORRUPTED;
		}
		map->m_flags |= EROFS_MAP_META;
	} else {
		erofs_err(sb, "internal error @ nid: %llu (size %llu), m_la 0x%llx",
			  vi->nid, inode->i_size, map->m_la);
		DBG_BUGON(1);
		return -EIO;
	}
	return 0;
}

int erofs_map_blocks(struct inode *inode, struct erofs_map_blocks *map)
{
	struct super_block *sb = inode->i_sb;
	struct erofs_inode *vi = EROFS_I(inode);
	struct erofs_inode_chunk_index *idx;
	struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
	u64 chunknr;
	unsigned int unit;
	erofs_off_t pos;
	void *kaddr;
	int err = 0;

	trace_erofs_map_blocks_enter(inode, map, 0);
	map->m_deviceid = 0;
	if (map->m_la >= inode->i_size) {
		/* leave out-of-bound access unmapped */
		map->m_flags = 0;
		map->m_plen = 0;
		goto out;
	}

	if (vi->datalayout != EROFS_INODE_CHUNK_BASED) {
		err = erofs_map_blocks_flatmode(inode, map);
		goto out;
	}

	if (vi->chunkformat & EROFS_CHUNK_FORMAT_INDEXES)
		unit = sizeof(*idx);			/* chunk index */
	else
		unit = EROFS_BLOCK_MAP_ENTRY_SIZE;	/* block map */

	chunknr = map->m_la >> vi->chunkbits;
	pos = ALIGN(erofs_iloc(inode) + vi->inode_isize +
		    vi->xattr_isize, unit) + unit * chunknr;

	kaddr = erofs_read_metabuf(&buf, sb, erofs_blknr(sb, pos), EROFS_KMAP);
	if (IS_ERR(kaddr)) {
		err = PTR_ERR(kaddr);
		goto out;
	}
	map->m_la = chunknr << vi->chunkbits;
	map->m_plen = min_t(erofs_off_t, 1UL << vi->chunkbits,
			round_up(inode->i_size - map->m_la, sb->s_blocksize));

	/* handle block map */
	if (!(vi->chunkformat & EROFS_CHUNK_FORMAT_INDEXES)) {
		__le32 *blkaddr = kaddr + erofs_blkoff(sb, pos);

		if (le32_to_cpu(*blkaddr) == EROFS_NULL_ADDR) {
			map->m_flags = 0;
		} else {
			map->m_pa = erofs_pos(sb, le32_to_cpu(*blkaddr));
			map->m_flags = EROFS_MAP_MAPPED;
		}
		goto out_unlock;
	}
	/* parse chunk indexes */
	idx = kaddr + erofs_blkoff(sb, pos);
	switch (le32_to_cpu(idx->blkaddr)) {
	case EROFS_NULL_ADDR:
		map->m_flags = 0;
		break;
	default:
		map->m_deviceid = le16_to_cpu(idx->device_id) &
			EROFS_SB(sb)->device_id_mask;
		map->m_pa = erofs_pos(sb, le32_to_cpu(idx->blkaddr));
		map->m_flags = EROFS_MAP_MAPPED;
		break;
	}
out_unlock:
	erofs_put_metabuf(&buf);
out:
	if (!err)
		map->m_llen = map->m_plen;
	trace_erofs_map_blocks_exit(inode, map, 0, err);
	return err;
}

int erofs_map_dev(struct super_block *sb, struct erofs_map_dev *map)
{
	struct erofs_dev_context *devs = EROFS_SB(sb)->devs;
	struct erofs_device_info *dif;
	int id;

	map->m_bdev = sb->s_bdev;
	map->m_daxdev = EROFS_SB(sb)->dax_dev;
	map->m_dax_part_off = EROFS_SB(sb)->dax_part_off;
	map->m_fscache = EROFS_SB(sb)->s_fscache;

	if (map->m_deviceid) {
		down_read(&devs->rwsem);
		dif = idr_find(&devs->tree, map->m_deviceid - 1);
		if (!dif) {
			up_read(&devs->rwsem);
			return -ENODEV;
		}
		if (devs->flatdev) {
			map->m_pa += erofs_pos(sb, dif->mapped_blkaddr);
			up_read(&devs->rwsem);
			return 0;
		}
		map->m_bdev = dif->bdev;
		map->m_daxdev = dif->dax_dev;
		map->m_dax_part_off = dif->dax_part_off;
		map->m_fscache = dif->fscache;
		up_read(&devs->rwsem);
	} else if (devs->extra_devices && !devs->flatdev) {
		down_read(&devs->rwsem);
		idr_for_each_entry(&devs->tree, dif, id) {
			erofs_off_t startoff, length;

			if (!dif->mapped_blkaddr)
				continue;
			startoff = erofs_pos(sb, dif->mapped_blkaddr);
			length = erofs_pos(sb, dif->blocks);

			if (map->m_pa >= startoff &&
			    map->m_pa < startoff + length) {
				map->m_pa -= startoff;
				map->m_bdev = dif->bdev;
				map->m_daxdev = dif->dax_dev;
				map->m_dax_part_off = dif->dax_part_off;
				map->m_fscache = dif->fscache;
				break;
			}
		}
		up_read(&devs->rwsem);
	}
	return 0;
}

static int erofs_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
		unsigned int flags, struct iomap *iomap, struct iomap *srcmap)
{
	int ret;
	struct super_block *sb = inode->i_sb;
	struct erofs_map_blocks map;
	struct erofs_map_dev mdev;

	map.m_la = offset;
	map.m_llen = length;

	ret = erofs_map_blocks(inode, &map);
	if (ret < 0)
		return ret;

	mdev = (struct erofs_map_dev) {
		.m_deviceid = map.m_deviceid,
		.m_pa = map.m_pa,
	};
	ret = erofs_map_dev(sb, &mdev);
	if (ret)
		return ret;

	iomap->offset = map.m_la;
	if (flags & IOMAP_DAX)
		iomap->dax_dev = mdev.m_daxdev;
	else
		iomap->bdev = mdev.m_bdev;
	iomap->length = map.m_llen;
	iomap->flags = 0;
	iomap->private = NULL;

	if (!(map.m_flags & EROFS_MAP_MAPPED)) {
		iomap->type = IOMAP_HOLE;
		iomap->addr = IOMAP_NULL_ADDR;
		if (!iomap->length)
			iomap->length = length;
		return 0;
	}

	if (map.m_flags & EROFS_MAP_META) {
		void *ptr;
		struct erofs_buf buf = __EROFS_BUF_INITIALIZER;

		iomap->type = IOMAP_INLINE;
		ptr = erofs_read_metabuf(&buf, sb,
				erofs_blknr(sb, mdev.m_pa), EROFS_KMAP);
		if (IS_ERR(ptr))
			return PTR_ERR(ptr);
		iomap->inline_data = ptr + erofs_blkoff(sb, mdev.m_pa);
		iomap->private = buf.base;
	} else {
		iomap->type = IOMAP_MAPPED;
		iomap->addr = mdev.m_pa;
		if (flags & IOMAP_DAX)
			iomap->addr += mdev.m_dax_part_off;
	}
	return 0;
}

static int erofs_iomap_end(struct inode *inode, loff_t pos, loff_t length,
		ssize_t written, unsigned int flags, struct iomap *iomap)
{
	void *ptr = iomap->private;

	if (ptr) {
		struct erofs_buf buf = {
			.page = kmap_to_page(ptr),
			.base = ptr,
			.kmap_type = EROFS_KMAP,
		};

		DBG_BUGON(iomap->type != IOMAP_INLINE);
		erofs_put_metabuf(&buf);
	} else {
		DBG_BUGON(iomap->type == IOMAP_INLINE);
	}
	return written;
}

static const struct iomap_ops erofs_iomap_ops = {
	.iomap_begin = erofs_iomap_begin,
	.iomap_end = erofs_iomap_end,
};

int erofs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
		 u64 start, u64 len)
{
	if (erofs_inode_is_data_compressed(EROFS_I(inode)->datalayout)) {
#ifdef CONFIG_EROFS_FS_ZIP
		return iomap_fiemap(inode, fieinfo, start, len,
				    &z_erofs_iomap_report_ops);
#else
		return -EOPNOTSUPP;
#endif
	}
	return iomap_fiemap(inode, fieinfo, start, len, &erofs_iomap_ops);
}

/*
 * since we dont have write or truncate flows, so no inode
 * locking needs to be held at the moment.
 */
static int erofs_read_folio(struct file *file, struct folio *folio)
{
	return iomap_read_folio(folio, &erofs_iomap_ops);
}

static void erofs_readahead(struct readahead_control *rac)
{
	return iomap_readahead(rac, &erofs_iomap_ops);
}

static sector_t erofs_bmap(struct address_space *mapping, sector_t block)
{
	return iomap_bmap(mapping, block, &erofs_iomap_ops);
}

static ssize_t erofs_file_read_iter(struct kiocb *iocb, struct iov_iter *to)
{
	struct inode *inode = file_inode(iocb->ki_filp);

	/* no need taking (shared) inode lock since it's a ro filesystem */
	if (!iov_iter_count(to))
		return 0;

#ifdef CONFIG_FS_DAX
	if (IS_DAX(inode))
		return dax_iomap_rw(iocb, to, &erofs_iomap_ops);
#endif
	if (iocb->ki_flags & IOCB_DIRECT) {
		struct block_device *bdev = inode->i_sb->s_bdev;
		unsigned int blksize_mask;

		if (bdev)
			blksize_mask = bdev_logical_block_size(bdev) - 1;
		else
			blksize_mask = i_blocksize(inode) - 1;

		if ((iocb->ki_pos | iov_iter_count(to) |
		     iov_iter_alignment(to)) & blksize_mask)
			return -EINVAL;

		return iomap_dio_rw(iocb, to, &erofs_iomap_ops,
				    NULL, 0, NULL, 0);
	}
	return filemap_read(iocb, to, 0);
}

/* for uncompressed (aligned) files and raw access for other files */
const struct address_space_operations erofs_raw_access_aops = {
	.read_folio = erofs_read_folio,
	.readahead = erofs_readahead,
	.bmap = erofs_bmap,
	.direct_IO = noop_direct_IO,
	.release_folio = iomap_release_folio,
	.invalidate_folio = iomap_invalidate_folio,
};

#ifdef CONFIG_FS_DAX
static vm_fault_t erofs_dax_huge_fault(struct vm_fault *vmf,
		enum page_entry_size pe_size)
{
	return dax_iomap_fault(vmf, pe_size, NULL, NULL, &erofs_iomap_ops);
}

static vm_fault_t erofs_dax_fault(struct vm_fault *vmf)
{
	return erofs_dax_huge_fault(vmf, PE_SIZE_PTE);
}

static const struct vm_operations_struct erofs_dax_vm_ops = {
	.fault		= erofs_dax_fault,
	.huge_fault	= erofs_dax_huge_fault,
};

static int erofs_file_mmap(struct file *file, struct vm_area_struct *vma)
{
	if (!IS_DAX(file_inode(file)))
		return generic_file_readonly_mmap(file, vma);

	if ((vma->vm_flags & VM_SHARED) && (vma->vm_flags & VM_MAYWRITE))
		return -EINVAL;

	vma->vm_ops = &erofs_dax_vm_ops;
	vm_flags_set(vma, VM_HUGEPAGE);
	return 0;
}
#else
#define erofs_file_mmap	generic_file_readonly_mmap
#endif

const struct file_operations erofs_file_fops = {
	.llseek		= generic_file_llseek,
	.read_iter	= erofs_file_read_iter,
	.mmap		= erofs_file_mmap,
	.splice_read	= filemap_splice_read,
};
Commit	Line	Data
29b24f6c	1	// SPDX-License-Identifier: GPL-2.0-only
81781b02	2	/*
81781b02	3	* Copyright (C) 2017-2018 HUAWEI, Inc.
592e7cd0	4	* https://www.huawei.com/
c5aa903a	5	* Copyright (C) 2021, Alibaba Cloud
81781b02 GX	6	*/
	7	#include "internal.h"
	8	#include <linux/prefetch.h>
5375e7c8	9	#include <linux/sched/mm.h>
06252e9c	10	#include <linux/dax.h>
13f06f48 CY	11	#include <trace/events/erofs.h>
13f06f48 CY	12
fdf80a47 GX	13	void erofs_unmap_metabuf(struct erofs_buf *buf)
	14	{
	15	if (buf->kmap_type == EROFS_KMAP)
927e5010	16	kunmap_local(buf->base);
fdf80a47 GX	17	buf->base = NULL;
	18	buf->kmap_type = EROFS_NO_KMAP;
	19	}
	20
	21	void erofs_put_metabuf(struct erofs_buf *buf)
	22	{
	23	if (!buf->page)
	24	return;
	25	erofs_unmap_metabuf(buf);
	26	put_page(buf->page);
	27	buf->page = NULL;
	28	}
	29
3acea5fc JX	30	/*
	31	* Derive the block size from inode->i_blkbits to make compatible with
	32	* anonymous inode in fscache mode.
	33	*/
eb2c5e41 GX	34	void erofs_bread(struct erofs_buf buf, erofs_blk_t blkaddr,
eb2c5e41 GX	35	enum erofs_kmap_type type)
fdf80a47	36	{
eb2c5e41	37	struct inode *inode = buf->inode;
3acea5fc	38	erofs_off_t offset = (erofs_off_t)blkaddr << inode->i_blkbits;
fdf80a47 GX	39	pgoff_t index = offset >> PAGE_SHIFT;
fdf80a47 GX	40	struct page *page = buf->page;
5375e7c8 JX	41	struct folio *folio;
5375e7c8 JX	42	unsigned int nofs_flag;
fdf80a47 GX	43
	44	if (!page \|\| page->index != index) {
	45	erofs_put_metabuf(buf);
5375e7c8 JX	46
5375e7c8 JX	47	nofs_flag = memalloc_nofs_save();
eb2c5e41	48	folio = read_cache_folio(inode->i_mapping, index, NULL, NULL);
5375e7c8 JX	49	memalloc_nofs_restore(nofs_flag);
	50	if (IS_ERR(folio))
	51	return folio;
	52
fdf80a47	53	/* should already be PageUptodate, no need to lock page */
5375e7c8	54	page = folio_file_page(folio, index);
fdf80a47 GX	55	buf->page = page;
	56	}
	57	if (buf->kmap_type == EROFS_NO_KMAP) {
	58	if (type == EROFS_KMAP)
927e5010	59	buf->base = kmap_local_page(page);
fdf80a47 GX	60	buf->kmap_type = type;
	61	} else if (buf->kmap_type != type) {
	62	DBG_BUGON(1);
	63	return ERR_PTR(-EFAULT);
	64	}
	65	if (type == EROFS_NO_KMAP)
	66	return NULL;
	67	return buf->base + (offset & ~PAGE_MASK);
	68	}
	69
eb2c5e41	70	void erofs_init_metabuf(struct erofs_buf buf, struct super_block sb)
fe5de585	71	{
5375e7c8	72	if (erofs_is_fscache_mode(sb))
eb2c5e41 GX	73	buf->inode = EROFS_SB(sb)->s_fscache->inode;
	74	else
	75	buf->inode = sb->s_bdev->bd_inode;
	76	}
5375e7c8	77
eb2c5e41 GX	78	void erofs_read_metabuf(struct erofs_buf buf, struct super_block *sb,
	79	erofs_blk_t blkaddr, enum erofs_kmap_type type)
	80	{
	81	erofs_init_metabuf(buf, sb);
	82	return erofs_bread(buf, blkaddr, type);
fe5de585 GX	83	}
fe5de585 GX	84
81781b02	85	static int erofs_map_blocks_flatmode(struct inode *inode,
8b58f9f0	86	struct erofs_map_blocks *map)
81781b02 GX	87	{
	88	erofs_blk_t nblocks, lastblk;
	89	u64 offset = map->m_la;
a5876e24	90	struct erofs_inode *vi = EROFS_I(inode);
3acea5fc	91	struct super_block *sb = inode->i_sb;
8a765682	92	bool tailendpacking = (vi->datalayout == EROFS_INODE_FLAT_INLINE);
81781b02	93
3acea5fc	94	nblocks = erofs_iblks(inode);
8a765682	95	lastblk = nblocks - tailendpacking;
81781b02	96
81781b02 GX	97	/* there is no hole in flatmode */
81781b02 GX	98	map->m_flags = EROFS_MAP_MAPPED;
3acea5fc JX	99	if (offset < erofs_pos(sb, lastblk)) {
	100	map->m_pa = erofs_pos(sb, vi->raw_blkaddr) + map->m_la;
	101	map->m_plen = erofs_pos(sb, lastblk) - offset;
8a765682	102	} else if (tailendpacking) {
b780d3fc	103	map->m_pa = erofs_iloc(inode) + vi->inode_isize +
3acea5fc	104	vi->xattr_isize + erofs_blkoff(sb, offset);
81781b02 GX	105	map->m_plen = inode->i_size - offset;
81781b02 GX	106
469407a3	107	/* inline data should be located in the same meta block */
3acea5fc JX	108	if (erofs_blkoff(sb, map->m_pa) + map->m_plen > sb->s_blocksize) {
3acea5fc JX	109	erofs_err(sb, "inline data cross block boundary @ nid %llu",
4f761fa2	110	vi->nid);
9141b60c	111	DBG_BUGON(1);
469407a3	112	return -EFSCORRUPTED;
9141b60c	113	}
81781b02 GX	114	map->m_flags \|= EROFS_MAP_META;
81781b02 GX	115	} else {
3acea5fc	116	erofs_err(sb, "internal error @ nid: %llu (size %llu), m_la 0x%llx",
4f761fa2	117	vi->nid, inode->i_size, map->m_la);
9141b60c	118	DBG_BUGON(1);
469407a3	119	return -EIO;
81781b02	120	}
469407a3	121	return 0;
81781b02 GX	122	}
81781b02 GX	123
8b58f9f0	124	int erofs_map_blocks(struct inode inode, struct erofs_map_blocks map)
c5aa903a GX	125	{
	126	struct super_block *sb = inode->i_sb;
	127	struct erofs_inode *vi = EROFS_I(inode);
	128	struct erofs_inode_chunk_index *idx;
fdf80a47	129	struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
c5aa903a GX	130	u64 chunknr;
	131	unsigned int unit;
	132	erofs_off_t pos;
fdf80a47	133	void *kaddr;
c5aa903a GX	134	int err = 0;
c5aa903a GX	135
8b58f9f0	136	trace_erofs_map_blocks_enter(inode, map, 0);
dfeab2e9	137	map->m_deviceid = 0;
c5aa903a GX	138	if (map->m_la >= inode->i_size) {
	139	/* leave out-of-bound access unmapped */
	140	map->m_flags = 0;
	141	map->m_plen = 0;
	142	goto out;
	143	}
	144
469407a3	145	if (vi->datalayout != EROFS_INODE_CHUNK_BASED) {
8b58f9f0	146	err = erofs_map_blocks_flatmode(inode, map);
469407a3 GX	147	goto out;
469407a3 GX	148	}
c5aa903a GX	149
	150	if (vi->chunkformat & EROFS_CHUNK_FORMAT_INDEXES)
	151	unit = sizeof(idx); / chunk index */
	152	else
	153	unit = EROFS_BLOCK_MAP_ENTRY_SIZE; /* block map */
	154
	155	chunknr = map->m_la >> vi->chunkbits;
b780d3fc	156	pos = ALIGN(erofs_iloc(inode) + vi->inode_isize +
c5aa903a GX	157	vi->xattr_isize, unit) + unit * chunknr;
c5aa903a GX	158
3acea5fc	159	kaddr = erofs_read_metabuf(&buf, sb, erofs_blknr(sb, pos), EROFS_KMAP);
fdf80a47 GX	160	if (IS_ERR(kaddr)) {
fdf80a47 GX	161	err = PTR_ERR(kaddr);
469407a3 GX	162	goto out;
469407a3 GX	163	}
c5aa903a GX	164	map->m_la = chunknr << vi->chunkbits;
c5aa903a GX	165	map->m_plen = min_t(erofs_off_t, 1UL << vi->chunkbits,
3acea5fc	166	round_up(inode->i_size - map->m_la, sb->s_blocksize));
c5aa903a GX	167
	168	/* handle block map */
	169	if (!(vi->chunkformat & EROFS_CHUNK_FORMAT_INDEXES)) {
3acea5fc	170	__le32 *blkaddr = kaddr + erofs_blkoff(sb, pos);
c5aa903a GX	171
	172	if (le32_to_cpu(*blkaddr) == EROFS_NULL_ADDR) {
	173	map->m_flags = 0;
	174	} else {
3acea5fc	175	map->m_pa = erofs_pos(sb, le32_to_cpu(*blkaddr));
c5aa903a GX	176	map->m_flags = EROFS_MAP_MAPPED;
	177	}
	178	goto out_unlock;
	179	}
	180	/* parse chunk indexes */
3acea5fc	181	idx = kaddr + erofs_blkoff(sb, pos);
c5aa903a GX	182	switch (le32_to_cpu(idx->blkaddr)) {
	183	case EROFS_NULL_ADDR:
	184	map->m_flags = 0;
	185	break;
	186	default:
dfeab2e9 GX	187	map->m_deviceid = le16_to_cpu(idx->device_id) &
dfeab2e9 GX	188	EROFS_SB(sb)->device_id_mask;
3acea5fc	189	map->m_pa = erofs_pos(sb, le32_to_cpu(idx->blkaddr));
c5aa903a GX	190	map->m_flags = EROFS_MAP_MAPPED;
	191	break;
	192	}
	193	out_unlock:
fdf80a47	194	erofs_put_metabuf(&buf);
c5aa903a	195	out:
469407a3 GX	196	if (!err)
469407a3 GX	197	map->m_llen = map->m_plen;
8b58f9f0	198	trace_erofs_map_blocks_exit(inode, map, 0, err);
c5aa903a GX	199	return err;
	200	}
	201
dfeab2e9 GX	202	int erofs_map_dev(struct super_block sb, struct erofs_map_dev map)
	203	{
	204	struct erofs_dev_context *devs = EROFS_SB(sb)->devs;
	205	struct erofs_device_info *dif;
	206	int id;
	207
dfeab2e9 GX	208	map->m_bdev = sb->s_bdev;
dfeab2e9 GX	209	map->m_daxdev = EROFS_SB(sb)->dax_dev;
de205114	210	map->m_dax_part_off = EROFS_SB(sb)->dax_part_off;
955b478e	211	map->m_fscache = EROFS_SB(sb)->s_fscache;
dfeab2e9 GX	212
	213	if (map->m_deviceid) {
	214	down_read(&devs->rwsem);
	215	dif = idr_find(&devs->tree, map->m_deviceid - 1);
	216	if (!dif) {
	217	up_read(&devs->rwsem);
	218	return -ENODEV;
	219	}
8b465fec JZ	220	if (devs->flatdev) {
	221	map->m_pa += erofs_pos(sb, dif->mapped_blkaddr);
	222	up_read(&devs->rwsem);
	223	return 0;
	224	}
dfeab2e9 GX	225	map->m_bdev = dif->bdev;
dfeab2e9 GX	226	map->m_daxdev = dif->dax_dev;
de205114	227	map->m_dax_part_off = dif->dax_part_off;
955b478e	228	map->m_fscache = dif->fscache;
dfeab2e9	229	up_read(&devs->rwsem);
8b465fec	230	} else if (devs->extra_devices && !devs->flatdev) {
dfeab2e9 GX	231	down_read(&devs->rwsem);
	232	idr_for_each_entry(&devs->tree, dif, id) {
	233	erofs_off_t startoff, length;
	234
	235	if (!dif->mapped_blkaddr)
	236	continue;
3acea5fc JX	237	startoff = erofs_pos(sb, dif->mapped_blkaddr);
3acea5fc JX	238	length = erofs_pos(sb, dif->blocks);
dfeab2e9 GX	239
	240	if (map->m_pa >= startoff &&
	241	map->m_pa < startoff + length) {
	242	map->m_pa -= startoff;
	243	map->m_bdev = dif->bdev;
	244	map->m_daxdev = dif->dax_dev;
de205114	245	map->m_dax_part_off = dif->dax_part_off;
955b478e	246	map->m_fscache = dif->fscache;
dfeab2e9 GX	247	break;
	248	}
	249	}
	250	up_read(&devs->rwsem);
	251	}
	252	return 0;
	253	}
	254
a08e67a0 HJ	255	static int erofs_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
	256	unsigned int flags, struct iomap iomap, struct iomap srcmap)
	257	{
	258	int ret;
3acea5fc	259	struct super_block *sb = inode->i_sb;
a08e67a0	260	struct erofs_map_blocks map;
dfeab2e9	261	struct erofs_map_dev mdev;
a08e67a0 HJ	262
	263	map.m_la = offset;
	264	map.m_llen = length;
	265
8b58f9f0	266	ret = erofs_map_blocks(inode, &map);
a08e67a0 HJ	267	if (ret < 0)
	268	return ret;
	269
dfeab2e9 GX	270	mdev = (struct erofs_map_dev) {
	271	.m_deviceid = map.m_deviceid,
	272	.m_pa = map.m_pa,
	273	};
3acea5fc	274	ret = erofs_map_dev(sb, &mdev);
dfeab2e9 GX	275	if (ret)
	276	return ret;
	277
a08e67a0	278	iomap->offset = map.m_la;
e33f42b2	279	if (flags & IOMAP_DAX)
de205114	280	iomap->dax_dev = mdev.m_daxdev;
e33f42b2	281	else
de205114	282	iomap->bdev = mdev.m_bdev;
a08e67a0 HJ	283	iomap->length = map.m_llen;
a08e67a0 HJ	284	iomap->flags = 0;
771c994e	285	iomap->private = NULL;
a08e67a0 HJ	286
	287	if (!(map.m_flags & EROFS_MAP_MAPPED)) {
	288	iomap->type = IOMAP_HOLE;
	289	iomap->addr = IOMAP_NULL_ADDR;
	290	if (!iomap->length)
	291	iomap->length = length;
	292	return 0;
	293	}
	294
a08e67a0	295	if (map.m_flags & EROFS_MAP_META) {
fdf80a47 GX	296	void *ptr;
fdf80a47 GX	297	struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
771c994e GX	298
771c994e GX	299	iomap->type = IOMAP_INLINE;
3acea5fc JX	300	ptr = erofs_read_metabuf(&buf, sb,
3acea5fc JX	301	erofs_blknr(sb, mdev.m_pa), EROFS_KMAP);
fdf80a47 GX	302	if (IS_ERR(ptr))
fdf80a47 GX	303	return PTR_ERR(ptr);
3acea5fc	304	iomap->inline_data = ptr + erofs_blkoff(sb, mdev.m_pa);
fdf80a47	305	iomap->private = buf.base;
771c994e GX	306	} else {
771c994e GX	307	iomap->type = IOMAP_MAPPED;
dfeab2e9	308	iomap->addr = mdev.m_pa;
e33f42b2 GX	309	if (flags & IOMAP_DAX)
e33f42b2 GX	310	iomap->addr += mdev.m_dax_part_off;
a08e67a0	311	}
a08e67a0 HJ	312	return 0;
	313	}
	314
771c994e GX	315	static int erofs_iomap_end(struct inode *inode, loff_t pos, loff_t length,
	316	ssize_t written, unsigned int flags, struct iomap *iomap)
	317	{
fdf80a47 GX	318	void *ptr = iomap->private;
	319
	320	if (ptr) {
	321	struct erofs_buf buf = {
	322	.page = kmap_to_page(ptr),
	323	.base = ptr,
	324	.kmap_type = EROFS_KMAP,
	325	};
771c994e	326
771c994e	327	DBG_BUGON(iomap->type != IOMAP_INLINE);
fdf80a47	328	erofs_put_metabuf(&buf);
771c994e GX	329	} else {
	330	DBG_BUGON(iomap->type == IOMAP_INLINE);
	331	}
	332	return written;
	333	}
	334
a08e67a0 HJ	335	static const struct iomap_ops erofs_iomap_ops = {
a08e67a0 HJ	336	.iomap_begin = erofs_iomap_begin,
771c994e	337	.iomap_end = erofs_iomap_end,
a08e67a0 HJ	338	};
a08e67a0 HJ	339
eadcd6b5 GX	340	int erofs_fiemap(struct inode inode, struct fiemap_extent_info fieinfo,
	341	u64 start, u64 len)
	342	{
	343	if (erofs_inode_is_data_compressed(EROFS_I(inode)->datalayout)) {
	344	#ifdef CONFIG_EROFS_FS_ZIP
	345	return iomap_fiemap(inode, fieinfo, start, len,
	346	&z_erofs_iomap_report_ops);
	347	#else
	348	return -EOPNOTSUPP;
	349	#endif
	350	}
	351	return iomap_fiemap(inode, fieinfo, start, len, &erofs_iomap_ops);
	352	}
	353
771c994e GX	354	/*
	355	* since we dont have write or truncate flows, so no inode
	356	* locking needs to be held at the moment.
	357	*/
7479c505	358	static int erofs_read_folio(struct file file, struct folio folio)
771c994e	359	{
7479c505	360	return iomap_read_folio(folio, &erofs_iomap_ops);
771c994e GX	361	}
	362
	363	static void erofs_readahead(struct readahead_control *rac)
	364	{
	365	return iomap_readahead(rac, &erofs_iomap_ops);
	366	}
	367
	368	static sector_t erofs_bmap(struct address_space *mapping, sector_t block)
	369	{
	370	return iomap_bmap(mapping, block, &erofs_iomap_ops);
	371	}
	372
de8a801a	373	static ssize_t erofs_file_read_iter(struct kiocb iocb, struct iov_iter to)
a08e67a0 HJ	374	{
a08e67a0 HJ	375	struct inode *inode = file_inode(iocb->ki_filp);
a08e67a0	376
a08e67a0 HJ	377	/* no need taking (shared) inode lock since it's a ro filesystem */
	378	if (!iov_iter_count(to))
	379	return 0;
	380
06252e9c	381	#ifdef CONFIG_FS_DAX
de8a801a	382	if (IS_DAX(inode))
06252e9c GX	383	return dax_iomap_rw(iocb, to, &erofs_iomap_ops);
06252e9c GX	384	#endif
a08e67a0	385	if (iocb->ki_flags & IOCB_DIRECT) {
de8a801a GX	386	struct block_device *bdev = inode->i_sb->s_bdev;
	387	unsigned int blksize_mask;
	388
	389	if (bdev)
	390	blksize_mask = bdev_logical_block_size(bdev) - 1;
	391	else
3993f4f4	392	blksize_mask = i_blocksize(inode) - 1;
de8a801a GX	393
	394	if ((iocb->ki_pos \| iov_iter_count(to) \|
	395	iov_iter_alignment(to)) & blksize_mask)
	396	return -EINVAL;
a08e67a0	397
de8a801a GX	398	return iomap_dio_rw(iocb, to, &erofs_iomap_ops,
de8a801a GX	399	NULL, 0, NULL, 0);
a08e67a0 HJ	400	}
	401	return filemap_read(iocb, to, 0);
	402	}
	403
81781b02 GX	404	/* for uncompressed (aligned) files and raw access for other files */
81781b02 GX	405	const struct address_space_operations erofs_raw_access_aops = {
7479c505	406	.read_folio = erofs_read_folio,
771c994e	407	.readahead = erofs_readahead,
9da681e0	408	.bmap = erofs_bmap,
a08e67a0	409	.direct_IO = noop_direct_IO,
ce529cc2 JX	410	.release_folio = iomap_release_folio,
ce529cc2 JX	411	.invalidate_folio = iomap_invalidate_folio,
a08e67a0 HJ	412	};
a08e67a0 HJ	413
06252e9c GX	414	#ifdef CONFIG_FS_DAX
	415	static vm_fault_t erofs_dax_huge_fault(struct vm_fault *vmf,
	416	enum page_entry_size pe_size)
	417	{
	418	return dax_iomap_fault(vmf, pe_size, NULL, NULL, &erofs_iomap_ops);
	419	}
	420
	421	static vm_fault_t erofs_dax_fault(struct vm_fault *vmf)
	422	{
	423	return erofs_dax_huge_fault(vmf, PE_SIZE_PTE);
	424	}
	425
	426	static const struct vm_operations_struct erofs_dax_vm_ops = {
	427	.fault = erofs_dax_fault,
	428	.huge_fault = erofs_dax_huge_fault,
	429	};
	430
	431	static int erofs_file_mmap(struct file file, struct vm_area_struct vma)
	432	{
	433	if (!IS_DAX(file_inode(file)))
	434	return generic_file_readonly_mmap(file, vma);
	435
	436	if ((vma->vm_flags & VM_SHARED) && (vma->vm_flags & VM_MAYWRITE))
	437	return -EINVAL;
	438
	439	vma->vm_ops = &erofs_dax_vm_ops;
1c71222e	440	vm_flags_set(vma, VM_HUGEPAGE);
06252e9c GX	441	return 0;
	442	}
	443	#else
	444	#define erofs_file_mmap generic_file_readonly_mmap
	445	#endif
	446
a08e67a0 HJ	447	const struct file_operations erofs_file_fops = {
	448	.llseek = generic_file_llseek,
	449	.read_iter = erofs_file_read_iter,
06252e9c	450	.mmap = erofs_file_mmap,
2cb1e089	451	.splice_read = filemap_splice_read,
81781b02	452	};