Merge tag 'i3c/for-7.3' of git://git.kernel.org/pub/scm/linux/kernel/git/i3c/linuxgrafted

Pull I3C updates from Alexandre Belloni:
 "A new driver this cycle, for the AMD AXI I3C controller.

  There are also new features such as IBI wakeup support, SETAASA device
  discovery and ACPI support for the the DAA methods, meaning we can now
  communicate with the SPD devices on DDR5 modules.

  New driver:
   - AMD AXI I3C master controller

  Subsystem:
   - I3C Common Command Code (CCC) handling improvements, especially
     around GET CCCs
   - SETAASA device discovery support
   - ACPI support for all existing DAA methods like SETDASA, SETNEWDA as
     well as I2C devices on I3C bus
   - IBI-based wakeup support

  Drivers:
   - dw: SETAASA support
   - mipi-i3c-hci: advertise IBI wakeup capability, AMD_PT I3C
     controller support, PIO queue management support for HCI v1.2
   - renesas: improve suspend to RAM support, add runtime PM support"

* tag 'i3c/for-7.3' of git://git.kernel.org/pub/scm/linux/kernel/git/i3c/linux: (70 commits)
  i3c: dw: reduce do_daa time if there's no client
  i3c: mipi-i3c-hci: Add support for AMD_PT I3C controller
  i3c: mipi-i3c-hci: Add PIO queue management support for HCI v1.2
  i3c: mipi-i3c-hci: Fix missing STAT_IBI_STATUS_THLD in PIO mode
  i3c: dw: rename "pclk" to "apb" to match dt-binding
  i3c: dw: make struct dw_i3c_cmd smaller
  i3c: dw: use COMMAND_PORT_TRANSFER_ARG instead of hardcoding
  i3c: renesas: Don't register devices when ENTDAA times out
  i3c: master: dw-i3c-master: fix OD timing for first broadcast
  i3c: mipi-i3c-hci: Advertise IBI wakeup capability
  i3c: mipi-i3c-hci: Factor out i3c_hci_sysdev()
  i3c: mipi-i3c-hci-pci: Propagate I3C wakeup requirements to PCI
  i3c: master: Reject IBI requests from non-IBI-capable devices
  i3c: master: Add helper to query bus wakeup requirements
  i3c: master: Report wakeup events for IBIs
  i3c: master: Support IBI-based wakeup capability
  i3c: master: Fix potential UAF in i3c_device_match()
  i3c: master: Fix potential UAF in i3c_device_uevent()
  i3c: Make dev->desc locking assumptions explicit
  i3c: master: Fix use-after-free of master->this
  ...

23 files changed, 9623 insertions, 0 deletions

diff --git a/fs/erofs/Kconfig b/fs/erofs/Kconfig
new file mode 100644
index 000000000..2dfc31358
--- /dev/null
+++ b/fs/erofs/Kconfig
@@ -0,0 +1,212 @@
+# SPDX-License-Identifier: GPL-2.0-only
+
+config EROFS_FS
+	tristate "EROFS filesystem support"
+	depends on BLOCK
+	select CRC32
+	select CRYPTO if EROFS_FS_ZIP_ACCEL
+	select CRYPTO_DEFLATE if EROFS_FS_ZIP_ACCEL
+	select FS_IOMAP
+	select LZ4_DECOMPRESS if EROFS_FS_ZIP
+	select XXHASH if EROFS_FS_XATTR
+	select XZ_DEC if EROFS_FS_ZIP_LZMA
+	select XZ_DEC_MICROLZMA if EROFS_FS_ZIP_LZMA
+	select ZLIB_INFLATE if EROFS_FS_ZIP_DEFLATE
+	select ZSTD_DECOMPRESS if EROFS_FS_ZIP_ZSTD
+	help
+	  EROFS (Enhanced Read-Only File System) is a modern, lightweight,
+	  secure read-only filesystem for various use cases, such as immutable
+	  system images, container images, application sandboxes, and datasets.
+
+	  EROFS uses a flexible, hierarchical on-disk design so that features
+	  can be enabled on demand: the core on-disk format is block-aligned in
+	  order to perform optimally on all kinds of devices, including block
+	  and memory-backed devices; the format is easy to parse and has zero
+	  metadata redundancy, unlike generic filesystems, making it ideal for
+	  filesystem auditing and remote access; inline data, random-access
+	  friendly directory data, inline/shared extended attributes and
+	  chunk-based deduplication ensure space efficiency while maintaining
+	  high performance.
+
+	  Optionally, it supports multiple devices to reference external data,
+	  enabling data sharing for container images.
+
+	  It also has advanced encoded on-disk layouts, particularly for data
+	  compression and fine-grained deduplication. It utilizes fixed-size
+	  output compression to improve storage density while keeping relatively
+	  high compression ratios. Furthermore, it implements in-place
+	  decompression to reuse file pages to keep compressed data temporarily
+	  with proper strategies, which ensures guaranteed end-to-end runtime
+	  performance under extreme memory pressure without extra cost.
+
+	  For more details, see the web pages at <https://erofs.docs.kernel.org>
+	  and the documentation at <file:Documentation/filesystems/erofs.rst>.
+
+	  To compile EROFS filesystem support as a module, choose M here. The
+	  module will be called erofs.
+
+	  If unsure, say N.
+
+config EROFS_FS_DEBUG
+	bool "EROFS debugging feature"
+	depends on EROFS_FS
+	help
+	  Print debugging messages and enable more BUG_ONs which check
+	  filesystem consistency and find potential issues aggressively,
+	  which can be used for Android eng build, for example.
+
+	  For daily use, say N.
+
+config EROFS_FS_XATTR
+	bool "EROFS extended attributes"
+	depends on EROFS_FS
+	default y
+	help
+	  Extended attributes are name:value pairs associated with inodes by
+	  the kernel or by users (see the attr(5) manual page, or visit
+	  <http://acl.bestbits.at/> for details).
+
+	  If unsure, say N.
+
+config EROFS_FS_POSIX_ACL
+	bool "EROFS Access Control Lists"
+	depends on EROFS_FS_XATTR
+	select FS_POSIX_ACL
+	default y
+	help
+	  Posix Access Control Lists (ACLs) support permissions for users and
+	  groups beyond the owner/group/world scheme.
+
+	  To learn more about Access Control Lists, visit the POSIX ACLs for
+	  Linux website <http://acl.bestbits.at/>.
+
+	  If you don't know what Access Control Lists are, say N.
+
+config EROFS_FS_SECURITY
+	bool "EROFS Security Labels"
+	depends on EROFS_FS_XATTR
+	default y
+	help
+	  Security labels provide an access control facility to support Linux
+	  Security Models (LSMs) accepted by AppArmor, SELinux, Smack and TOMOYO
+	  Linux. This option enables an extended attribute handler for file
+	  security labels in the erofs filesystem, so that it requires enabling
+	  the extended attribute support in advance.
+
+	  If you are not using a security module, say N.
+
+config EROFS_FS_BACKED_BY_FILE
+	bool "File-backed EROFS filesystem support"
+	depends on EROFS_FS
+	default y
+	help
+	  This allows EROFS to use filesystem image files directly, without
+	  the intercession of loopback block devices or likewise. It is
+	  particularly useful for container images with numerous blobs and
+	  other sandboxes, where loop devices behave intricately.  It can also
+	  be used to simplify error-prone lifetime management of unnecessary
+	  virtual block devices.
+
+	  If you don't want to enable this feature, say N.
+
+config EROFS_FS_ZIP
+	bool "EROFS Data Compression Support"
+	depends on EROFS_FS
+	default y
+	help
+	  Enable EROFS compression layouts so that filesystems containing
+	  compressed files can be parsed by the kernel.
+
+	  If you don't want to enable compression feature, say N.
+
+config EROFS_FS_ZIP_LZMA
+	bool "EROFS LZMA compressed data support"
+	depends on EROFS_FS_ZIP
+	default y
+	help
+	  Saying Y here includes support for reading EROFS file systems
+	  containing LZMA compressed data, specifically called microLZMA. It
+	  gives better compression ratios than the default LZ4 format, at the
+	  expense of more CPU overhead.
+
+	  Say N if you want to disable LZMA compression support.
+
+config EROFS_FS_ZIP_LZMA_DEFAULT_MAX_STREAMS
+	int "EROFS LZMA default maximum decompression streams"
+	depends on EROFS_FS_ZIP_LZMA
+	range 1 NR_CPUS if SMP
+	range 1 1 if !SMP
+	default 16
+	help
+	  By default EROFS allocates one LZMA decompression stream per CPU.
+	  Each stream can hold a dictionary of up to 8 MiB taken from the
+	  mounted image, so on systems with many CPUs this can reserve a lot
+	  of memory.  This caps the default; the lzma_streams module parameter
+	  still overrides it.
+
+	  If unsure, keep the default of 16.
+
+config EROFS_FS_ZIP_DEFLATE
+	bool "EROFS DEFLATE compressed data support"
+	depends on EROFS_FS_ZIP
+	help
+	  Saying Y here includes support for reading EROFS file systems
+	  containing DEFLATE compressed data.  It gives better compression
+	  ratios than the default LZ4 format, while it costs more CPU
+	  overhead.
+
+	  If unsure, say N.
+
+config EROFS_FS_ZIP_ZSTD
+	bool "EROFS Zstandard compressed data support"
+	depends on EROFS_FS_ZIP
+	help
+	  Saying Y here includes support for reading EROFS file systems
+	  containing Zstandard compressed data.  It gives better compression
+	  ratios than the default LZ4 format, while it costs more CPU
+	  overhead and memory footprint.
+
+	  If unsure, say N.
+
+config EROFS_FS_ZIP_ACCEL
+	bool "EROFS hardware decompression support"
+	depends on EROFS_FS_ZIP
+	help
+	  Saying Y here includes hardware accelerator support for reading
+	  EROFS file systems containing compressed data.  It gives better
+	  decompression speed than the software-implemented decompression, and
+	  it costs lower CPU overhead.
+
+	  Hardware accelerator support is an experimental feature for now and
+	  file systems are still readable without selecting this option.
+
+	  If unsure, say N.
+
+config EROFS_FS_PCPU_KTHREAD
+	bool "EROFS per-cpu decompression kthread workers"
+	depends on EROFS_FS_ZIP
+	help
+	  Saying Y here enables per-CPU kthread workers pool to carry out
+	  async decompression for low latencies on some architectures.
+
+	  If unsure, say N.
+
+config EROFS_FS_PCPU_KTHREAD_HIPRI
+	bool "EROFS high priority per-CPU kthread workers"
+	depends on EROFS_FS_ZIP && EROFS_FS_PCPU_KTHREAD
+	default y
+	help
+	  This permits EROFS to configure per-CPU kthread workers to run
+	  at higher priority.
+
+	  If unsure, say N.
+
+config EROFS_FS_PAGE_CACHE_SHARE
+	bool "EROFS page cache share support (experimental)"
+	depends on EROFS_FS && EROFS_FS_XATTR
+	select FS_STACK
+	help
+	  This enables page cache sharing among inodes with identical
+	  content fingerprints on the same machine.
+
+	  If unsure, say N.
diff --git a/fs/erofs/Makefile b/fs/erofs/Makefile
new file mode 100644
index 000000000..304234967
--- /dev/null
+++ b/fs/erofs/Makefile
@@ -0,0 +1,12 @@
+# SPDX-License-Identifier: GPL-2.0-only
+
+obj-$(CONFIG_EROFS_FS) += erofs.o
+erofs-objs := super.o inode.o data.o namei.o dir.o sysfs.o
+erofs-$(CONFIG_EROFS_FS_XATTR) += xattr.o
+erofs-$(CONFIG_EROFS_FS_ZIP) += decompressor.o zmap.o zdata.o zutil.o
+erofs-$(CONFIG_EROFS_FS_ZIP_LZMA) += decompressor_lzma.o
+erofs-$(CONFIG_EROFS_FS_ZIP_DEFLATE) += decompressor_deflate.o
+erofs-$(CONFIG_EROFS_FS_ZIP_ZSTD) += decompressor_zstd.o
+erofs-$(CONFIG_EROFS_FS_ZIP_ACCEL) += decompressor_crypto.o
+erofs-$(CONFIG_EROFS_FS_BACKED_BY_FILE) += fileio.o
+erofs-$(CONFIG_EROFS_FS_PAGE_CACHE_SHARE) += ishare.o
diff --git a/fs/erofs/compress.h b/fs/erofs/compress.h
new file mode 100644
index 000000000..84c8e5258
--- /dev/null
+++ b/fs/erofs/compress.h
@@ -0,0 +1,89 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (C) 2019 HUAWEI, Inc.
+ *             https://www.huawei.com/
+ */
+#ifndef __EROFS_FS_COMPRESS_H
+#define __EROFS_FS_COMPRESS_H
+
+#include "internal.h"
+
+struct z_erofs_decompress_req {
+	struct super_block *sb;
+	struct page **in, **out;
+	unsigned int inpages, outpages;
+	unsigned short pageofs_in, pageofs_out;
+	unsigned int inputsize, outputsize;
+
+	unsigned int alg;       /* the algorithm for decompression */
+	bool inplace_io, partial_decoding, fillgaps;
+	gfp_t gfp;      /* allocation flags for extra temporary buffers */
+};
+
+struct z_erofs_decompressor {
+	int (*config)(struct super_block *sb, struct erofs_super_block *dsb,
+		      void *data, int size);
+	const char *(*decompress)(struct z_erofs_decompress_req *rq,
+				  struct page **pagepool);
+	int (*init)(void);
+	void (*exit)(void);
+	char *name;
+};
+
+#define Z_EROFS_SHORTLIVED_PAGE		(-1UL << 2)
+#define Z_EROFS_PREALLOCATED_FOLIO	((void *)(-2UL << 2))
+
+/*
+ * Currently, short-lived pages are pages directly from buddy system
+ * with specific page->private (Z_EROFS_SHORTLIVED_PAGE).
+ * In the future world of Memdescs, it should be type 0 (Misc) memory
+ * which type can be checked with a new helper.
+ */
+static inline bool z_erofs_is_shortlived_page(struct page *page)
+{
+	return page->private == Z_EROFS_SHORTLIVED_PAGE;
+}
+
+static inline bool z_erofs_put_shortlivedpage(struct page **pagepool,
+					      struct page *page)
+{
+	if (!z_erofs_is_shortlived_page(page))
+		return false;
+	erofs_pagepool_add(pagepool, page);
+	return true;
+}
+
+extern const struct z_erofs_decompressor z_erofs_lzma_decomp;
+extern const struct z_erofs_decompressor z_erofs_deflate_decomp;
+extern const struct z_erofs_decompressor z_erofs_zstd_decomp;
+extern const struct z_erofs_decompressor *z_erofs_decomp[];
+
+struct z_erofs_stream_dctx {
+	struct z_erofs_decompress_req *rq;
+	int no, ni;			/* the current {en,de}coded page # */
+
+	unsigned int avail_out;		/* remaining bytes in the decoded buffer */
+	unsigned int inbuf_pos, inbuf_sz;
+					/* current status of the encoded buffer */
+	u8 *kin, *kout;			/* buffer mapped pointers */
+	void *bounce;			/* bounce buffer for inplace I/Os */
+	bool bounced;			/* is the bounce buffer used now? */
+};
+
+const char *z_erofs_stream_switch_bufs(struct z_erofs_stream_dctx *dctx,
+				void **dst, void **src, struct page **pgpl);
+const char *z_erofs_fixup_insize(struct z_erofs_decompress_req *rq,
+				 const char *padbuf, unsigned int padbufsize);
+int __init z_erofs_init_decompressor(void);
+void z_erofs_exit_decompressor(void);
+int z_erofs_crypto_decompress(struct z_erofs_decompress_req *rq,
+			      struct page **pgpl);
+int z_erofs_crypto_enable_engine(const char *name, int len);
+#ifdef CONFIG_EROFS_FS_ZIP_ACCEL
+void z_erofs_crypto_disable_all_engines(void);
+int z_erofs_crypto_show_engines(char *buf, int size, char sep);
+#else
+static inline void z_erofs_crypto_disable_all_engines(void) {}
+static inline int z_erofs_crypto_show_engines(char *buf, int size, char sep) { return 0; }
+#endif
+#endif
diff --git a/fs/erofs/data.c b/fs/erofs/data.c
new file mode 100644
index 000000000..20962f8b6
--- /dev/null
+++ b/fs/erofs/data.c
@@ -0,0 +1,534 @@
+// 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/filelock.h>
+#include <linux/sched/mm.h>
+#include <trace/events/erofs.h>
+
+void erofs_unmap_metabuf(struct erofs_buf *buf)
+{
+	if (!buf->base)
+		return;
+	kunmap_local(buf->base);
+	buf->base = NULL;
+}
+
+void erofs_put_metabuf(struct erofs_buf *buf)
+{
+	if (!buf->page)
+		return;
+	erofs_unmap_metabuf(buf);
+	folio_put(page_folio(buf->page));
+	buf->page = NULL;
+}
+
+void *erofs_bread(struct erofs_buf *buf, erofs_off_t offset, bool need_kmap)
+{
+	pgoff_t index = (buf->off + offset) >> PAGE_SHIFT;
+	struct folio *folio = NULL;
+
+	if (buf->page) {
+		folio = page_folio(buf->page);
+		if (folio_file_page(folio, index) != buf->page)
+			erofs_unmap_metabuf(buf);
+	}
+	if (!folio || !folio_contains(folio, index)) {
+		erofs_put_metabuf(buf);
+		folio = read_cache_folio(buf->mapping, index,
+				buf->mc ? erofs_read_meta_folio : NULL, NULL);
+		if (IS_ERR(folio))
+			return folio;
+	}
+	buf->page = folio_file_page(folio, index);
+	if (!need_kmap)
+		return NULL;
+	if (!buf->base)
+		buf->base = kmap_local_page(buf->page);
+	return buf->base + (offset & ~PAGE_MASK);
+}
+
+int erofs_init_metabuf(struct erofs_buf *buf, struct super_block *sb,
+		       bool in_metabox)
+{
+	struct erofs_sb_info *sbi = EROFS_SB(sb);
+
+	buf->mc = false;
+	if (in_metabox) {
+		if (unlikely(!sbi->metabox_inode))
+			return -EFSCORRUPTED;
+		buf->mapping = sbi->metabox_inode->i_mapping;
+		return 0;
+	}
+	if (erofs_is_fileio_mode(sbi)) {
+		buf->mapping = sbi->managed_cache->i_mapping;
+		buf->mc = true;
+	} else {
+		buf->off = sbi->dif0.fsoff;
+		buf->mapping = sb->s_bdev->bd_mapping;
+	}
+	return 0;
+}
+
+void *erofs_read_metabuf(struct erofs_buf *buf, struct super_block *sb,
+			 erofs_off_t offset, bool in_metabox)
+{
+	int err;
+
+	err = erofs_init_metabuf(buf, sb, in_metabox);
+	if (err)
+		return ERR_PTR(err);
+	return erofs_bread(buf, offset, true);
+}
+
+static int erofs_map_chunks(struct inode *inode, struct erofs_map_blocks *map)
+{
+	struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
+	struct super_block *sb = inode->i_sb;
+	struct erofs_inode *vi = EROFS_I(inode);
+	struct erofs_inode_chunk_index *idx;
+	unsigned int unit = vi->chunkformat & EROFS_CHUNK_FORMAT_INDEXES ?
+		sizeof(*idx) : EROFS_BLOCK_MAP_ENTRY_SIZE;
+	erofs_blk_t addrmask = (vi->chunkformat & EROFS_CHUNK_FORMAT_48BIT) ?
+		BIT_ULL(48) - 1 : BIT_ULL(32) - 1;
+	u64 nr = map->m_la >> vi->chunkbits, chunksize = 1ULL << vi->chunkbits;
+	erofs_off_t pos = ALIGN(erofs_iloc(inode) + vi->inode_isize +
+				vi->xattr_isize, unit) + unit * nr;
+	/* m_llen will be clamped to EOF in the end */
+	erofs_off_t endpos = round_up(pos + 1, sb->s_blocksize);
+	u64 last, addr;
+
+	idx = erofs_read_metabuf(&buf, sb, pos, erofs_inode_in_metabox(inode));
+	if (IS_ERR(idx))
+		return PTR_ERR(idx);
+
+	map->m_la = nr << vi->chunkbits;
+	map->m_llen = 0;
+	nr = 0;
+	do {
+		if (unit == EROFS_BLOCK_MAP_ENTRY_SIZE) {
+			addr = le32_to_cpu(((__le32 *)idx)[nr]);
+			if (addr == (u32)EROFS_NULL_ADDR)
+				addr = EROFS_NULL_ADDR;
+		} else {
+			addr = (((u64)le16_to_cpu(idx[nr].startblk_hi) << 32) |
+				le32_to_cpu(idx[nr].startblk_lo)) & addrmask;
+			if (addr ^ (EROFS_NULL_ADDR & addrmask))
+				addr |= (u64)(le16_to_cpu(idx[nr].device_id) &
+					EROFS_SB(sb)->device_id_mask) << 48;
+			else
+				addr = EROFS_NULL_ADDR;
+		}
+		if (!nr) {
+			last = addr;
+			continue;
+		}
+		/* expand and account the prior chunk here */
+		map->m_llen += chunksize;
+		if (last != EROFS_NULL_ADDR)
+			last += erofs_blknr(sb, chunksize);
+	} while (addr == last && pos + (++nr) * unit < endpos);
+
+	if (last != EROFS_NULL_ADDR) {
+		map->m_pa = erofs_pos(sb, last & addrmask) - map->m_llen;
+		map->m_deviceid = last >> 48;
+		map->m_flags = EROFS_MAP_MAPPED;
+	}
+	if (addr == last)
+		map->m_llen += chunksize;
+	map->m_llen = min_t(erofs_off_t, map->m_llen,
+			round_up(inode->i_size - map->m_la, sb->s_blocksize));
+	erofs_put_metabuf(&buf);
+	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);
+	bool tailinline = (vi->datalayout == EROFS_INODE_FLAT_INLINE);
+	erofs_off_t pos;
+	int err = 0;
+
+	trace_erofs_map_blocks_enter(inode, map, 0);
+	map->m_deviceid = 0;
+	map->m_flags = 0;
+	if (map->m_la >= inode->i_size)
+		goto out;
+	if (vi->datalayout == EROFS_INODE_CHUNK_BASED) {
+		err = erofs_map_chunks(inode, map);
+	} else if (tailinline || vi->startblk != EROFS_NULL_ADDR) {
+		pos = erofs_pos(sb, erofs_iblks(inode) - tailinline);
+		map->m_flags = EROFS_MAP_MAPPED;
+		if (map->m_la < pos) {
+			map->m_pa = erofs_pos(sb, vi->startblk) + map->m_la;
+			map->m_llen = pos - map->m_la;
+		} else {
+			map->m_pa = erofs_iloc(inode) + vi->inode_isize +
+				vi->xattr_isize + erofs_blkoff(sb, map->m_la);
+			map->m_llen = inode->i_size - map->m_la;
+			map->m_flags |= EROFS_MAP_META;
+			if (erofs_blkoff(sb, map->m_pa) + map->m_llen >
+			    sb->s_blocksize) {
+				erofs_err(sb, "inline data across blocks @ nid %llu", vi->nid);
+				return -EFSCORRUPTED;
+			}
+		}
+	}
+out:
+	map->m_plen = err ? 0 : map->m_llen;
+	trace_erofs_map_blocks_exit(inode, map, 0, err);
+	return err;
+}
+
+static void erofs_fill_from_devinfo(struct erofs_map_dev *map,
+		struct super_block *sb, struct erofs_device_info *dif)
+{
+	map->m_sb = sb;
+	map->m_dif = dif;
+	map->m_bdev = NULL;
+	if (dif->file && S_ISBLK(file_inode(dif->file)->i_mode))
+		map->m_bdev = file_bdev(dif->file);
+}
+
+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;
+	erofs_off_t startoff;
+	int id;
+
+	erofs_fill_from_devinfo(map, sb, &EROFS_SB(sb)->dif0);
+	map->m_bdev = sb->s_bdev;	/* use s_bdev for the primary device */
+	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->uniaddr);
+			up_read(&devs->rwsem);
+			return 0;
+		}
+		erofs_fill_from_devinfo(map, sb, dif);
+		up_read(&devs->rwsem);
+	} else if (devs->extra_devices && !devs->flatdev) {
+		down_read(&devs->rwsem);
+		idr_for_each_entry(&devs->tree, dif, id) {
+			if (!dif->uniaddr)
+				continue;
+
+			startoff = erofs_pos(sb, dif->uniaddr);
+			if (map->m_pa >= startoff &&
+			    map->m_pa < startoff + erofs_pos(sb, dif->blocks)) {
+				map->m_pa -= startoff;
+				erofs_fill_from_devinfo(map, sb, dif);
+				break;
+			}
+		}
+		up_read(&devs->rwsem);
+	}
+	return 0;
+}
+
+/*
+ * bit 30: I/O error occurred on this folio
+ * bit 29: CPU has dirty data in D-cache (needs aliasing handling);
+ * bit 0 - 29: remaining parts to complete this folio
+ */
+#define EROFS_ONLINEFOLIO_EIO		30
+#define EROFS_ONLINEFOLIO_DIRTY		29
+
+void erofs_onlinefolio_init(struct folio *folio)
+{
+	union {
+		atomic_t o;
+		void *v;
+	} u = { .o = ATOMIC_INIT(1) };
+
+	folio->private = u.v;	/* valid only if file-backed folio is locked */
+}
+
+void erofs_onlinefolio_split(struct folio *folio)
+{
+	atomic_inc((atomic_t *)&folio->private);
+}
+
+void erofs_onlinefolio_end(struct folio *folio, int err, bool dirty)
+{
+	int orig, v;
+
+	do {
+		orig = atomic_read((atomic_t *)&folio->private);
+		DBG_BUGON(orig <= 0);
+		v = dirty << EROFS_ONLINEFOLIO_DIRTY;
+		v |= (orig - 1) | (!!err << EROFS_ONLINEFOLIO_EIO);
+	} while (atomic_cmpxchg((atomic_t *)&folio->private, orig, v) != orig);
+
+	if (v & (BIT(EROFS_ONLINEFOLIO_DIRTY) - 1))
+		return;
+	folio->private = 0;
+	if (v & BIT(EROFS_ONLINEFOLIO_DIRTY))
+		flush_dcache_folio(folio);
+	folio_end_read(folio, !(v & BIT(EROFS_ONLINEFOLIO_EIO)));
+}
+
+struct erofs_iomap_iter_ctx {
+	struct page *page;
+	void *base;
+	struct inode *realinode;
+};
+
+static int erofs_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
+		unsigned int flags, struct iomap *iomap, struct iomap *srcmap)
+{
+	struct iomap_iter *iter = container_of(iomap, struct iomap_iter, iomap);
+	struct erofs_iomap_iter_ctx *ctx = iter->private;
+	struct inode *realinode = ctx ? ctx->realinode : inode;
+	struct super_block *sb = realinode->i_sb;
+	struct erofs_map_blocks map;
+	struct erofs_map_dev mdev;
+	int ret;
+
+	map.m_la = offset;
+	map.m_llen = length;
+	ret = erofs_map_blocks(realinode, &map);
+	if (ret < 0)
+		return ret;
+
+	iomap->offset = map.m_la;
+	iomap->length = map.m_llen;
+	iomap->flags = 0;
+	iomap->addr = IOMAP_NULL_ADDR;
+	if (!(map.m_flags & EROFS_MAP_MAPPED)) {
+		iomap->type = IOMAP_HOLE;
+		return 0;
+	}
+
+	if (!(map.m_flags & EROFS_MAP_META) || !erofs_inode_in_metabox(realinode)) {
+		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;
+
+		if (flags & IOMAP_DAX)
+			iomap->dax_dev = mdev.m_dif->dax_dev;
+		else
+			iomap->bdev = mdev.m_bdev;
+		iomap->addr = mdev.m_dif->fsoff + mdev.m_pa;
+		if (flags & IOMAP_DAX)
+			iomap->addr += mdev.m_dif->dax_part_off;
+	}
+
+	if (map.m_flags & EROFS_MAP_META) {
+		iomap->type = IOMAP_INLINE;
+		/* read context should read the inlined data */
+		if (ctx) {
+			struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
+			void *ptr;
+
+			ptr = erofs_read_metabuf(&buf, sb, map.m_pa,
+						 erofs_inode_in_metabox(realinode));
+			if (IS_ERR(ptr))
+				return PTR_ERR(ptr);
+			iomap->inline_data = ptr;
+			ctx->page = buf.page;
+			ctx->base = buf.base;
+		}
+	} else {
+		iomap->type = IOMAP_MAPPED;
+	}
+	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)
+{
+	struct iomap_iter *iter = container_of(iomap, struct iomap_iter, iomap);
+	struct erofs_iomap_iter_ctx *ctx = iter->private;
+
+	if (ctx && ctx->base) {
+		struct erofs_buf buf = {
+			.page = ctx->page,
+			.base = ctx->base,
+		};
+
+		DBG_BUGON(iomap->type != IOMAP_INLINE);
+		erofs_put_metabuf(&buf);
+		ctx->base = NULL;
+	}
+	return written;
+}
+
+static DEFINE_IOMAP_ITER_NEXT_END(erofs_iomap_next, erofs_iomap_begin,
+				  erofs_iomap_end);
+
+static const struct iomap_ops erofs_iomap_ops = {
+	.iomap_next = erofs_iomap_next,
+};
+
+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)) {
+		if (!IS_ENABLED(CONFIG_EROFS_FS_ZIP))
+			return -EOPNOTSUPP;
+		return iomap_fiemap(inode, fieinfo, start, len,
+				    &z_erofs_iomap_report_ops);
+	}
+	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)
+{
+	struct iomap_read_folio_ctx read_ctx = {
+		.ops		= &iomap_bio_read_ops,
+		.cur_folio	= folio,
+	};
+	bool need_iput;
+	struct erofs_iomap_iter_ctx iter_ctx = {
+		.realinode = erofs_real_inode(folio_inode(folio), &need_iput),
+	};
+
+	trace_erofs_read_folio(iter_ctx.realinode, folio, true);
+	iomap_read_folio(&erofs_iomap_ops, &read_ctx, &iter_ctx);
+	if (need_iput)
+		iput(iter_ctx.realinode);
+	return 0;
+}
+
+static void erofs_readahead(struct readahead_control *rac)
+{
+	struct iomap_read_folio_ctx read_ctx = {
+		.ops		= &iomap_bio_read_ops,
+		.rac		= rac,
+	};
+	bool need_iput;
+	struct erofs_iomap_iter_ctx iter_ctx = {
+		.realinode = erofs_real_inode(rac->mapping->host, &need_iput),
+	};
+
+	trace_erofs_readahead(iter_ctx.realinode, readahead_index(rac),
+			      readahead_count(rac), true);
+	iomap_readahead(&erofs_iomap_ops, &read_ctx, &iter_ctx);
+	if (need_iput)
+		iput(iter_ctx.realinode);
+}
+
+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;
+
+	if (IS_ENABLED(CONFIG_FS_DAX) && IS_DAX(inode))
+		return dax_iomap_rw(iocb, to, &erofs_iomap_ops);
+
+	if ((iocb->ki_flags & IOCB_DIRECT) && inode->i_sb->s_bdev) {
+		struct erofs_iomap_iter_ctx iter_ctx = {
+			.realinode = inode,
+		};
+
+		return iomap_dio_rw(iocb, to, &erofs_iomap_ops,
+				    NULL, 0, &iter_ctx, 0);
+	}
+	return filemap_read(iocb, to, 0);
+}
+
+/* for uncompressed (aligned) files and raw access for other files */
+const struct address_space_operations erofs_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,
+		unsigned int order)
+{
+	return dax_iomap_fault(vmf, order, NULL, NULL, &erofs_iomap_ops);
+}
+
+static vm_fault_t erofs_dax_fault(struct vm_fault *vmf)
+{
+	return erofs_dax_huge_fault(vmf, 0);
+}
+
+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_prepare(struct vm_area_desc *desc)
+{
+	if (!IS_DAX(file_inode(desc->file)))
+		return generic_file_readonly_mmap_prepare(desc);
+
+	if (vma_desc_test_all(desc, VMA_SHARED_BIT, VMA_MAYWRITE_BIT))
+		return -EINVAL;
+
+	desc->vm_ops = &erofs_dax_vm_ops;
+	vma_desc_set_flags(desc, VMA_HUGEPAGE_BIT);
+	return 0;
+}
+#else
+#define erofs_file_mmap_prepare	generic_file_readonly_mmap_prepare
+#endif
+
+static loff_t erofs_file_llseek(struct file *file, loff_t offset, int whence)
+{
+	struct inode *inode = file->f_mapping->host;
+	const struct iomap_ops *ops = &erofs_iomap_ops;
+
+	if (erofs_inode_is_data_compressed(EROFS_I(inode)->datalayout)) {
+		if (!IS_ENABLED(CONFIG_EROFS_FS_ZIP))
+			return generic_file_llseek(file, offset, whence);
+		ops = &z_erofs_iomap_report_ops;
+	}
+
+	if (whence == SEEK_HOLE)
+		offset = iomap_seek_hole(inode, offset, ops);
+	else if (whence == SEEK_DATA)
+		offset = iomap_seek_data(inode, offset, ops);
+	else
+		return generic_file_llseek(file, offset, whence);
+
+	if (offset < 0)
+		return offset;
+	return vfs_setpos(file, offset, inode->i_sb->s_maxbytes);
+}
+
+const struct file_operations erofs_file_fops = {
+	.llseek		= erofs_file_llseek,
+	.read_iter	= erofs_file_read_iter,
+	.unlocked_ioctl = erofs_ioctl,
+#ifdef CONFIG_COMPAT
+	.compat_ioctl   = erofs_compat_ioctl,
+#endif
+	.mmap_prepare	= erofs_file_mmap_prepare,
+	.get_unmapped_area = thp_get_unmapped_area,
+	.splice_read	= filemap_splice_read,
+	.setlease	= generic_setlease,
+};
diff --git a/fs/erofs/decompressor.c b/fs/erofs/decompressor.c
new file mode 100644
index 000000000..27caf4beb
--- /dev/null
+++ b/fs/erofs/decompressor.c
@@ -0,0 +1,517 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2019 HUAWEI, Inc.
+ *             https://www.huawei.com/
+ * Copyright (C) 2024 Alibaba Cloud
+ */
+#include "compress.h"
+#include <linux/lz4.h>
+
+#define LZ4_MAX_DISTANCE_PAGES	(DIV_ROUND_UP(LZ4_DISTANCE_MAX, PAGE_SIZE) + 1)
+
+static int z_erofs_load_lz4_config(struct super_block *sb,
+			    struct erofs_super_block *dsb, void *data, int size)
+{
+	struct erofs_sb_info *sbi = EROFS_SB(sb);
+	struct z_erofs_lz4_cfgs *lz4 = data;
+	u16 distance;
+
+	if (lz4) {
+		if (size < sizeof(struct z_erofs_lz4_cfgs)) {
+			erofs_err(sb, "invalid lz4 cfgs, size=%u", size);
+			return -EINVAL;
+		}
+		distance = le16_to_cpu(lz4->max_distance);
+
+		sbi->lz4.max_pclusterblks = le16_to_cpu(lz4->max_pclusterblks);
+		if (!sbi->lz4.max_pclusterblks) {
+			sbi->lz4.max_pclusterblks = 1;	/* reserved case */
+		} else if (sbi->lz4.max_pclusterblks >
+			   erofs_blknr(sb, Z_EROFS_PCLUSTER_MAX_SIZE)) {
+			erofs_err(sb, "too large lz4 pclusterblks %u",
+				  sbi->lz4.max_pclusterblks);
+			return -EINVAL;
+		}
+	} else {
+		distance = le16_to_cpu(dsb->u1.lz4_max_distance);
+		if (!distance && !erofs_sb_has_lz4_0padding(sbi))
+			return 0;
+		sbi->lz4.max_pclusterblks = 1;
+		sbi->available_compr_algs = 1 << Z_EROFS_COMPRESSION_LZ4;
+	}
+
+	sbi->lz4.max_distance_pages = distance ?
+					DIV_ROUND_UP(distance, PAGE_SIZE) + 1 :
+					LZ4_MAX_DISTANCE_PAGES;
+	return z_erofs_gbuf_growsize(sbi->lz4.max_pclusterblks);
+}
+
+/*
+ * Fill all gaps with bounce pages if it's a sparse page list. Also check if
+ * all physical pages are consecutive, which can be seen for moderate CR.
+ */
+static int z_erofs_lz4_prepare_dstpages(struct z_erofs_decompress_req *rq,
+					struct page **pagepool)
+{
+	struct page *availables[LZ4_MAX_DISTANCE_PAGES] = { NULL };
+	unsigned long bounced[DIV_ROUND_UP(LZ4_MAX_DISTANCE_PAGES,
+					   BITS_PER_LONG)] = { 0 };
+	unsigned int lz4_max_distance_pages =
+				EROFS_SB(rq->sb)->lz4.max_distance_pages;
+	void *kaddr = NULL;
+	unsigned int i, j, top;
+
+	top = 0;
+	for (i = j = 0; i < rq->outpages; ++i, ++j) {
+		struct page *const page = rq->out[i];
+		struct page *victim;
+
+		if (j >= lz4_max_distance_pages)
+			j = 0;
+
+		/* 'valid' bounced can only be tested after a complete round */
+		if (!rq->fillgaps && test_bit(j, bounced)) {
+			DBG_BUGON(i < lz4_max_distance_pages);
+			DBG_BUGON(top >= lz4_max_distance_pages);
+			availables[top++] = rq->out[i - lz4_max_distance_pages];
+		}
+
+		if (page) {
+			__clear_bit(j, bounced);
+			if (!PageHighMem(page)) {
+				if (!i) {
+					kaddr = page_address(page);
+					continue;
+				}
+				if (kaddr &&
+				    kaddr + PAGE_SIZE == page_address(page)) {
+					kaddr += PAGE_SIZE;
+					continue;
+				}
+			}
+			kaddr = NULL;
+			continue;
+		}
+		kaddr = NULL;
+		__set_bit(j, bounced);
+
+		if (top) {
+			victim = availables[--top];
+		} else {
+			victim = __erofs_allocpage(pagepool, rq->gfp, true);
+			if (!victim)
+				return -ENOMEM;
+			set_page_private(victim, Z_EROFS_SHORTLIVED_PAGE);
+		}
+		rq->out[i] = victim;
+	}
+	return kaddr ? 1 : 0;
+}
+
+static void *z_erofs_lz4_handle_overlap(const struct z_erofs_decompress_req *rq,
+			void *inpage, void *out, unsigned int *inputmargin,
+			int *maptype, bool may_inplace)
+{
+	unsigned int oend, omargin, cnt, i;
+	struct page **in;
+	void *src;
+
+	/*
+	 * If in-place I/O isn't used, for example, the bounce compressed cache
+	 * can hold data for incomplete read requests. Just map the compressed
+	 * buffer as well and decompress directly.
+	 */
+	if (!rq->inplace_io) {
+		if (rq->inpages <= 1) {
+			*maptype = 0;
+			return inpage;
+		}
+		kunmap_local(inpage);
+		src = erofs_vm_map_ram(rq->in, rq->inpages);
+		if (!src)
+			return ERR_PTR(-ENOMEM);
+		*maptype = 1;
+		return src;
+	}
+	/*
+	 * Then, deal with in-place I/Os. The reasons why in-place I/O is useful
+	 * are: (1) It minimizes memory footprint during the I/O submission,
+	 * which is useful for slow storage (including network devices and
+	 * low-end HDDs/eMMCs) but with a lot inflight I/Os; (2) If in-place
+	 * decompression can also be applied, it will reuse the unique buffer so
+	 * that no extra CPU D-cache is polluted with temporary compressed data
+	 * for extreme performance.
+	 */
+	oend = rq->pageofs_out + rq->outputsize;
+	omargin = PAGE_ALIGN(oend) - oend;
+	if (!rq->partial_decoding && may_inplace &&
+	    omargin >= LZ4_DECOMPRESS_INPLACE_MARGIN(rq->inputsize)) {
+		for (i = 0; i < rq->inpages; ++i)
+			if (rq->out[rq->outpages - rq->inpages + i] !=
+			    rq->in[i])
+				break;
+		if (i >= rq->inpages) {
+			kunmap_local(inpage);
+			*maptype = 3;
+			return out + ((rq->outpages - rq->inpages) << PAGE_SHIFT);
+		}
+	}
+	/*
+	 * If in-place decompression can't be applied, copy compressed data that
+	 * may potentially overlap during decompression to a per-CPU buffer.
+	 */
+	src = z_erofs_get_gbuf(rq->inpages);
+	if (!src) {
+		DBG_BUGON(1);
+		kunmap_local(inpage);
+		return ERR_PTR(-EFAULT);
+	}
+
+	for (i = 0, in = rq->in; i < rq->inputsize; i += cnt, ++in) {
+		cnt = min_t(u32, rq->inputsize - i, PAGE_SIZE - *inputmargin);
+		if (!inpage)
+			inpage = kmap_local_page(*in);
+		memcpy(src + i, inpage + *inputmargin, cnt);
+		kunmap_local(inpage);
+		inpage = NULL;
+		*inputmargin = 0;
+	}
+	*maptype = 2;
+	return src;
+}
+
+/*
+ * Get the exact on-disk size of the compressed data:
+ *  - For LZ4, it should apply if the zero_padding feature is on (5.3+);
+ *  - For others, zero_padding is enabled all the time.
+ */
+const char *z_erofs_fixup_insize(struct z_erofs_decompress_req *rq,
+				 const char *padbuf, unsigned int padbufsize)
+{
+	const char *padend;
+
+	padend = memchr_inv(padbuf, 0, padbufsize);
+	if (!padend)
+		return "compressed data start not found";
+	rq->inputsize -= padend - padbuf;
+	rq->pageofs_in += padend - padbuf;
+	return NULL;
+}
+
+static const char *__z_erofs_lz4_decompress(struct z_erofs_decompress_req *rq,
+					    u8 *dst)
+{
+	bool may_inplace = false;
+	unsigned int inputmargin;
+	u8 *out, *headpage, *src;
+	const char *reason;
+	int ret, maptype;
+
+	headpage = kmap_local_page(*rq->in);
+	reason = z_erofs_fixup_insize(rq, headpage + rq->pageofs_in,
+			min_t(unsigned int, rq->inputsize,
+			      rq->sb->s_blocksize - rq->pageofs_in));
+	if (reason) {
+		kunmap_local(headpage);
+		return reason;
+	}
+	may_inplace = !((rq->pageofs_in + rq->inputsize) &
+			(rq->sb->s_blocksize - 1));
+
+	inputmargin = rq->pageofs_in;
+	src = z_erofs_lz4_handle_overlap(rq, headpage, dst, &inputmargin,
+					 &maptype, may_inplace);
+	if (IS_ERR(src))
+		return ERR_CAST(src);
+
+	out = dst + rq->pageofs_out;
+	if (rq->partial_decoding)
+		ret = LZ4_decompress_safe_partial(src + inputmargin, out,
+				rq->inputsize, rq->outputsize, rq->outputsize);
+	else
+		ret = LZ4_decompress_safe(src + inputmargin, out,
+					  rq->inputsize, rq->outputsize);
+	if (ret == rq->outputsize)
+		reason = NULL;
+	else if (ret < 0)
+		reason = "corrupted compressed data";
+	else
+		reason = "unexpected end of stream";
+
+	if (!maptype) {
+		kunmap_local(headpage);
+	} else if (maptype == 1) {
+		vm_unmap_ram(src, rq->inpages);
+	} else if (maptype == 2) {
+		z_erofs_put_gbuf(src);
+	} else if (maptype != 3) {
+		DBG_BUGON(1);
+		return ERR_PTR(-EFAULT);
+	}
+	return reason;
+}
+
+static const char *z_erofs_lz4_decompress(struct z_erofs_decompress_req *rq,
+					  struct page **pagepool)
+{
+	unsigned int dst_maptype;
+	const char *reason;
+	void *dst;
+	int ret;
+
+	/* one optimized fast path only for non bigpcluster cases yet */
+	if (rq->inpages == 1 && rq->outpages == 1 && !rq->inplace_io) {
+		DBG_BUGON(!*rq->out);
+		dst = kmap_local_page(*rq->out);
+		dst_maptype = 0;
+	} else {
+		/* general decoding path which can be used for all cases */
+		ret = z_erofs_lz4_prepare_dstpages(rq, pagepool);
+		if (ret < 0)
+			return ERR_PTR(ret);
+		if (ret > 0) {
+			dst = page_address(*rq->out);
+			dst_maptype = 1;
+		} else {
+			dst = erofs_vm_map_ram(rq->out, rq->outpages);
+			if (!dst)
+				return ERR_PTR(-ENOMEM);
+			dst_maptype = 2;
+		}
+	}
+	reason = __z_erofs_lz4_decompress(rq, dst);
+	if (!dst_maptype)
+		kunmap_local(dst);
+	else if (dst_maptype == 2)
+		vm_unmap_ram(dst, rq->outpages);
+	return reason;
+}
+
+static const char *z_erofs_transform_plain(struct z_erofs_decompress_req *rq,
+					   struct page **pagepool)
+{
+	const unsigned int nrpages_in = rq->inpages, nrpages_out = rq->outpages;
+	const unsigned int bs = rq->sb->s_blocksize;
+	unsigned int cur = 0, ni = 0, no, pi, po, insz, cnt;
+	u8 *kin;
+
+	if (rq->outputsize > rq->inputsize)
+		return ERR_PTR(-EOPNOTSUPP);
+	if (rq->alg == Z_EROFS_COMPRESSION_INTERLACED) {
+		cur = bs - (rq->pageofs_out & (bs - 1));
+		DBG_BUGON(rq->pageofs_in & (bs - 1));
+		pi = (rq->pageofs_in + rq->inputsize - cur) & ~PAGE_MASK;
+		cur = min(cur, rq->outputsize);
+		if (cur && rq->out[0]) {
+			kin = kmap_local_page(rq->in[nrpages_in - 1]);
+			if (rq->out[0] == rq->in[nrpages_in - 1])
+				memmove(kin + rq->pageofs_out, kin + pi, cur);
+			else
+				memcpy_to_page(rq->out[0], rq->pageofs_out,
+					       kin + pi, cur);
+			kunmap_local(kin);
+		}
+		rq->outputsize -= cur;
+	}
+
+	for (; rq->outputsize; rq->pageofs_in = 0, cur += insz, ni++) {
+		insz = min(PAGE_SIZE - rq->pageofs_in, rq->outputsize);
+		rq->outputsize -= insz;
+		if (!rq->in[ni])
+			continue;
+		kin = kmap_local_page(rq->in[ni]);
+		pi = 0;
+		do {
+			no = (rq->pageofs_out + cur + pi) >> PAGE_SHIFT;
+			po = (rq->pageofs_out + cur + pi) & ~PAGE_MASK;
+			DBG_BUGON(no >= nrpages_out);
+			cnt = min(insz - pi, PAGE_SIZE - po);
+			if (rq->out[no] == rq->in[ni])
+				memmove(kin + po,
+					kin + rq->pageofs_in + pi, cnt);
+			else if (rq->out[no])
+				memcpy_to_page(rq->out[no], po,
+					       kin + rq->pageofs_in + pi, cnt);
+			pi += cnt;
+		} while (pi < insz);
+		kunmap_local(kin);
+	}
+	DBG_BUGON(ni > nrpages_in);
+	return NULL;
+}
+
+const char *z_erofs_stream_switch_bufs(struct z_erofs_stream_dctx *dctx,
+				void **dst, void **src, struct page **pgpl)
+{
+	struct z_erofs_decompress_req *rq = dctx->rq;
+	struct page **pgo, *tmppage;
+	unsigned int j;
+
+	if (!dctx->avail_out) {
+		if (++dctx->no >= rq->outpages || !rq->outputsize)
+			return "insufficient space for decompressed data";
+
+		if (dctx->kout)
+			kunmap_local(dctx->kout);
+		dctx->avail_out = min(rq->outputsize, PAGE_SIZE - rq->pageofs_out);
+		rq->outputsize -= dctx->avail_out;
+		pgo = &rq->out[dctx->no];
+		if (!*pgo && rq->fillgaps) {		/* deduped */
+			*pgo = erofs_allocpage(pgpl, rq->gfp);
+			if (!*pgo) {
+				dctx->kout = NULL;
+				return ERR_PTR(-ENOMEM);
+			}
+			set_page_private(*pgo, Z_EROFS_SHORTLIVED_PAGE);
+		}
+		if (*pgo) {
+			dctx->kout = kmap_local_page(*pgo);
+			*dst = dctx->kout + rq->pageofs_out;
+		} else {
+			*dst = dctx->kout = NULL;
+		}
+		rq->pageofs_out = 0;
+	}
+
+	if (dctx->inbuf_pos == dctx->inbuf_sz && rq->inputsize) {
+		if (++dctx->ni >= rq->inpages)
+			return "invalid compressed data";
+		if (dctx->kout) /* unlike kmap(), take care of the orders */
+			kunmap_local(dctx->kout);
+		kunmap_local(dctx->kin);
+
+		dctx->inbuf_sz = min_t(u32, rq->inputsize, PAGE_SIZE);
+		rq->inputsize -= dctx->inbuf_sz;
+		dctx->kin = kmap_local_page(rq->in[dctx->ni]);
+		*src = dctx->kin;
+		dctx->bounced = false;
+		if (dctx->kout) {
+			j = (u8 *)*dst - dctx->kout;
+			dctx->kout = kmap_local_page(rq->out[dctx->no]);
+			*dst = dctx->kout + j;
+		}
+		dctx->inbuf_pos = 0;
+	}
+
+	/*
+	 * Handle overlapping: Use the given bounce buffer if the input data is
+	 * under processing; Or utilize short-lived pages from the on-stack page
+	 * pool, where pages are shared among the same request.  Note that only
+	 * a few inplace I/O pages need to be doubled.
+	 */
+	if (!dctx->bounced && rq->out[dctx->no] == rq->in[dctx->ni]) {
+		memcpy(dctx->bounce, *src, dctx->inbuf_sz);
+		*src = dctx->bounce;
+		dctx->bounced = true;
+	}
+
+	for (j = dctx->ni + 1; j < rq->inpages; ++j) {
+		if (rq->out[dctx->no] != rq->in[j])
+			continue;
+		tmppage = erofs_allocpage(pgpl, rq->gfp);
+		if (!tmppage)
+			return ERR_PTR(-ENOMEM);
+		set_page_private(tmppage, Z_EROFS_SHORTLIVED_PAGE);
+		copy_highpage(tmppage, rq->in[j]);
+		rq->in[j] = tmppage;
+	}
+	return NULL;
+}
+
+const struct z_erofs_decompressor *z_erofs_decomp[] = {
+	[Z_EROFS_COMPRESSION_SHIFTED] = &(const struct z_erofs_decompressor) {
+		.decompress = z_erofs_transform_plain,
+		.name = "shifted"
+	},
+	[Z_EROFS_COMPRESSION_INTERLACED] = &(const struct z_erofs_decompressor) {
+		.decompress = z_erofs_transform_plain,
+		.name = "interlaced"
+	},
+	[Z_EROFS_COMPRESSION_LZ4] = &(const struct z_erofs_decompressor) {
+		.config = z_erofs_load_lz4_config,
+		.decompress = z_erofs_lz4_decompress,
+		.init = z_erofs_gbuf_init,
+		.exit = z_erofs_gbuf_exit,
+		.name = "lz4"
+	},
+#ifdef CONFIG_EROFS_FS_ZIP_LZMA
+	[Z_EROFS_COMPRESSION_LZMA] = &z_erofs_lzma_decomp,
+#endif
+#ifdef CONFIG_EROFS_FS_ZIP_DEFLATE
+	[Z_EROFS_COMPRESSION_DEFLATE] = &z_erofs_deflate_decomp,
+#endif
+#ifdef CONFIG_EROFS_FS_ZIP_ZSTD
+	[Z_EROFS_COMPRESSION_ZSTD] = &z_erofs_zstd_decomp,
+#endif
+};
+
+int z_erofs_parse_cfgs(struct super_block *sb, struct erofs_super_block *dsb)
+{
+	struct erofs_sb_info *sbi = EROFS_SB(sb);
+	struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
+	unsigned long algs, alg;
+	erofs_off_t offset;
+	int size, ret = 0;
+
+	if (!erofs_sb_has_compr_cfgs(sbi))
+		return z_erofs_load_lz4_config(sb, dsb, NULL, 0);
+
+	algs = le16_to_cpu(dsb->u1.available_compr_algs);
+	sbi->available_compr_algs = algs;
+	if (algs & ~Z_EROFS_ALL_COMPR_ALGS) {
+		erofs_err(sb, "unidentified algorithms %lx, please upgrade kernel",
+			  algs & ~Z_EROFS_ALL_COMPR_ALGS);
+		return -EOPNOTSUPP;
+	}
+
+	(void)erofs_init_metabuf(&buf, sb, false);
+	offset = EROFS_SUPER_OFFSET + sbi->sb_size;
+	for_each_set_bit(alg, &algs, Z_EROFS_COMPRESSION_MAX) {
+		const struct z_erofs_decompressor *dec = z_erofs_decomp[alg];
+		void *data;
+
+		data = erofs_read_metadata(sb, &buf, &offset, &size);
+		if (IS_ERR(data)) {
+			ret = PTR_ERR(data);
+			break;
+		}
+
+		if (dec && dec->config) {
+			ret = dec->config(sb, dsb, data, size);
+		} else {
+			erofs_err(sb, "algorithm %ld isn't enabled on this kernel",
+				  alg);
+			ret = -EOPNOTSUPP;
+		}
+		kfree(data);
+		if (ret)
+			break;
+	}
+	erofs_put_metabuf(&buf);
+	return ret;
+}
+
+int __init z_erofs_init_decompressor(void)
+{
+	int i, err;
+
+	for (i = 0; i < Z_EROFS_COMPRESSION_MAX; ++i) {
+		err = z_erofs_decomp[i] ? z_erofs_decomp[i]->init() : 0;
+		if (err) {
+			while (i--)
+				if (z_erofs_decomp[i])
+					z_erofs_decomp[i]->exit();
+			return err;
+		}
+	}
+	return 0;
+}
+
+void z_erofs_exit_decompressor(void)
+{
+	int i;
+
+	for (i = 0; i < Z_EROFS_COMPRESSION_MAX; ++i)
+		if (z_erofs_decomp[i])
+			z_erofs_decomp[i]->exit();
+}
diff --git a/fs/erofs/decompressor_crypto.c b/fs/erofs/decompressor_crypto.c
new file mode 100644
index 000000000..77c6bd535
--- /dev/null
+++ b/fs/erofs/decompressor_crypto.c
@@ -0,0 +1,182 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+#include <linux/scatterlist.h>
+#include <crypto/acompress.h>
+#include "compress.h"
+
+static int __z_erofs_crypto_decompress(struct z_erofs_decompress_req *rq,
+				       struct crypto_acomp *tfm)
+{
+	struct sg_table st_src, st_dst;
+	struct acomp_req *req;
+	struct crypto_wait wait;
+	const char *reason;
+	u8 *headpage;
+	int ret;
+
+	headpage = kmap_local_page(*rq->in);
+	reason = z_erofs_fixup_insize(rq, headpage + rq->pageofs_in,
+				min_t(unsigned int, rq->inputsize,
+				      rq->sb->s_blocksize - rq->pageofs_in));
+	kunmap_local(headpage);
+	if (reason)
+		return IS_ERR(reason) ? PTR_ERR(reason) : -EFSCORRUPTED;
+
+	req = acomp_request_alloc(tfm);
+	if (!req)
+		return -ENOMEM;
+
+	ret = sg_alloc_table_from_pages_segment(&st_src, rq->in, rq->inpages,
+			rq->pageofs_in, rq->inputsize, UINT_MAX, GFP_KERNEL);
+	if (ret < 0)
+		goto failed_src_alloc;
+
+	ret = sg_alloc_table_from_pages_segment(&st_dst, rq->out, rq->outpages,
+			rq->pageofs_out, rq->outputsize, UINT_MAX, GFP_KERNEL);
+	if (ret < 0)
+		goto failed_dst_alloc;
+
+	acomp_request_set_params(req, st_src.sgl,
+				 st_dst.sgl, rq->inputsize, rq->outputsize);
+
+	crypto_init_wait(&wait);
+	acomp_request_set_callback(req, CRYPTO_TFM_REQ_MAY_BACKLOG,
+				   crypto_req_done, &wait);
+
+	ret = crypto_wait_req(crypto_acomp_decompress(req), &wait);
+	if (ret) {
+		erofs_err(rq->sb, "failed to decompress %d in[%u, %u] out[%u]",
+			  ret, rq->inputsize, rq->pageofs_in, rq->outputsize);
+		ret = -EIO;
+	}
+
+	sg_free_table(&st_dst);
+failed_dst_alloc:
+	sg_free_table(&st_src);
+failed_src_alloc:
+	acomp_request_free(req);
+	return ret;
+}
+
+struct z_erofs_crypto_engine {
+	char *crypto_name;
+	struct crypto_acomp *tfm;
+};
+
+static struct z_erofs_crypto_engine *z_erofs_crypto[Z_EROFS_COMPRESSION_MAX] = {
+	[Z_EROFS_COMPRESSION_LZ4] = (struct z_erofs_crypto_engine[]) {
+		{},
+	},
+	[Z_EROFS_COMPRESSION_LZMA] = (struct z_erofs_crypto_engine[]) {
+		{},
+	},
+	[Z_EROFS_COMPRESSION_DEFLATE] = (struct z_erofs_crypto_engine[]) {
+		{ .crypto_name = "qat_deflate", },
+		{},
+	},
+	[Z_EROFS_COMPRESSION_ZSTD] = (struct z_erofs_crypto_engine[]) {
+		{},
+	},
+};
+static DECLARE_RWSEM(z_erofs_crypto_rwsem);
+
+static struct crypto_acomp *z_erofs_crypto_get_engine(int alg)
+{
+	struct z_erofs_crypto_engine *e;
+
+	for (e = z_erofs_crypto[alg]; e->crypto_name; ++e)
+		if (e->tfm)
+			return e->tfm;
+	return NULL;
+}
+
+int z_erofs_crypto_decompress(struct z_erofs_decompress_req *rq,
+			      struct page **pgpl)
+{
+	struct crypto_acomp *tfm;
+	int i, err;
+
+	down_read(&z_erofs_crypto_rwsem);
+	tfm = z_erofs_crypto_get_engine(rq->alg);
+	if (!tfm) {
+		err = -EOPNOTSUPP;
+		goto out;
+	}
+
+	for (i = 0; i < rq->outpages; i++) {
+		struct page *const page = rq->out[i];
+		struct page *victim;
+
+		if (!page) {
+			victim = __erofs_allocpage(pgpl, rq->gfp, true);
+			if (!victim) {
+				err = -ENOMEM;
+				goto out;
+			}
+			set_page_private(victim, Z_EROFS_SHORTLIVED_PAGE);
+			rq->out[i] = victim;
+		}
+	}
+	err = __z_erofs_crypto_decompress(rq, tfm);
+out:
+	up_read(&z_erofs_crypto_rwsem);
+	return err;
+}
+
+int z_erofs_crypto_enable_engine(const char *name, int len)
+{
+	struct z_erofs_crypto_engine *e;
+	struct crypto_acomp *tfm;
+	int alg;
+
+	down_write(&z_erofs_crypto_rwsem);
+	for (alg = 0; alg < Z_EROFS_COMPRESSION_MAX; ++alg) {
+		for (e = z_erofs_crypto[alg]; e->crypto_name; ++e) {
+			if (!strncmp(name, e->crypto_name, len)) {
+				if (e->tfm)
+					break;
+				tfm = crypto_alloc_acomp(e->crypto_name, 0, 0);
+				if (IS_ERR(tfm)) {
+					up_write(&z_erofs_crypto_rwsem);
+					return -EOPNOTSUPP;
+				}
+				e->tfm = tfm;
+				break;
+			}
+		}
+	}
+	up_write(&z_erofs_crypto_rwsem);
+	return 0;
+}
+
+void z_erofs_crypto_disable_all_engines(void)
+{
+	struct z_erofs_crypto_engine *e;
+	int alg;
+
+	down_write(&z_erofs_crypto_rwsem);
+	for (alg = 0; alg < Z_EROFS_COMPRESSION_MAX; ++alg) {
+		for (e = z_erofs_crypto[alg]; e->crypto_name; ++e) {
+			if (!e->tfm)
+				continue;
+			crypto_free_acomp(e->tfm);
+			e->tfm = NULL;
+		}
+	}
+	up_write(&z_erofs_crypto_rwsem);
+}
+
+int z_erofs_crypto_show_engines(char *buf, int size, char sep)
+{
+	struct z_erofs_crypto_engine *e;
+	int alg, len = 0;
+
+	for (alg = 0; alg < Z_EROFS_COMPRESSION_MAX; ++alg) {
+		for (e = z_erofs_crypto[alg]; e->crypto_name; ++e) {
+			if (!e->tfm)
+				continue;
+			len += scnprintf(buf + len, size - len, "%s%c",
+					 e->crypto_name, sep);
+		}
+	}
+	return len;
+}
diff --git a/fs/erofs/decompressor_deflate.c b/fs/erofs/decompressor_deflate.c
new file mode 100644
index 000000000..6167af1ff
--- /dev/null
+++ b/fs/erofs/decompressor_deflate.c
@@ -0,0 +1,203 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+#include <linux/zlib.h>
+#include "compress.h"
+
+struct z_erofs_deflate {
+	struct z_erofs_deflate *next;
+	struct z_stream_s z;
+	u8 bounce[PAGE_SIZE];
+};
+
+static DEFINE_SPINLOCK(z_erofs_deflate_lock);
+static unsigned int z_erofs_deflate_nstrms, z_erofs_deflate_avail_strms;
+static struct z_erofs_deflate *z_erofs_deflate_head;
+static DECLARE_WAIT_QUEUE_HEAD(z_erofs_deflate_wq);
+
+module_param_named(deflate_streams, z_erofs_deflate_nstrms, uint, 0444);
+
+static void z_erofs_deflate_exit(void)
+{
+	/* there should be no running fs instance */
+	while (z_erofs_deflate_avail_strms) {
+		struct z_erofs_deflate *strm;
+
+		spin_lock(&z_erofs_deflate_lock);
+		strm = z_erofs_deflate_head;
+		if (!strm) {
+			spin_unlock(&z_erofs_deflate_lock);
+			continue;
+		}
+		z_erofs_deflate_head = NULL;
+		spin_unlock(&z_erofs_deflate_lock);
+
+		while (strm) {
+			struct z_erofs_deflate *n = strm->next;
+
+			vfree(strm->z.workspace);
+			kfree(strm);
+			--z_erofs_deflate_avail_strms;
+			strm = n;
+		}
+	}
+}
+
+static int __init z_erofs_deflate_init(void)
+{
+	/* by default, use # of possible CPUs instead */
+	if (!z_erofs_deflate_nstrms)
+		z_erofs_deflate_nstrms = num_possible_cpus();
+	return 0;
+}
+
+static int z_erofs_load_deflate_config(struct super_block *sb,
+			struct erofs_super_block *dsb, void *data, int size)
+{
+	struct z_erofs_deflate_cfgs *dfl = data;
+	static DEFINE_MUTEX(deflate_resize_mutex);
+	static bool inited;
+
+	if (!dfl || size < sizeof(struct z_erofs_deflate_cfgs)) {
+		erofs_err(sb, "invalid deflate cfgs, size=%u", size);
+		return -EINVAL;
+	}
+
+	if (dfl->windowbits > MAX_WBITS) {
+		erofs_err(sb, "unsupported windowbits %u", dfl->windowbits);
+		return -EOPNOTSUPP;
+	}
+	mutex_lock(&deflate_resize_mutex);
+	if (!inited) {
+		for (; z_erofs_deflate_avail_strms < z_erofs_deflate_nstrms;
+		     ++z_erofs_deflate_avail_strms) {
+			struct z_erofs_deflate *strm;
+
+			strm = kzalloc_obj(*strm);
+			if (!strm)
+				goto failed;
+			/* XXX: in-kernel zlib cannot customize windowbits */
+			strm->z.workspace = vmalloc(zlib_inflate_workspacesize());
+			if (!strm->z.workspace) {
+				kfree(strm);
+				goto failed;
+			}
+
+			spin_lock(&z_erofs_deflate_lock);
+			strm->next = z_erofs_deflate_head;
+			z_erofs_deflate_head = strm;
+			spin_unlock(&z_erofs_deflate_lock);
+		}
+		inited = true;
+	}
+	mutex_unlock(&deflate_resize_mutex);
+	return 0;
+failed:
+	mutex_unlock(&deflate_resize_mutex);
+	z_erofs_deflate_exit();
+	return -ENOMEM;
+}
+
+static const char *__z_erofs_deflate_decompress(struct z_erofs_decompress_req *rq,
+						struct page **pgpl)
+{
+	struct super_block *sb = rq->sb;
+	struct z_erofs_stream_dctx dctx = { .rq = rq, .no = -1, .ni = 0 };
+	struct z_erofs_deflate *strm;
+	const char *reason;
+	int zerr;
+
+	/* 1. get the exact DEFLATE compressed size */
+	dctx.kin = kmap_local_page(*rq->in);
+	reason = z_erofs_fixup_insize(rq, dctx.kin + rq->pageofs_in,
+			min(rq->inputsize, sb->s_blocksize - rq->pageofs_in));
+	if (reason) {
+		kunmap_local(dctx.kin);
+		return reason;
+	}
+
+	/* 2. get an available DEFLATE context */
+again:
+	spin_lock(&z_erofs_deflate_lock);
+	strm = z_erofs_deflate_head;
+	if (!strm) {
+		spin_unlock(&z_erofs_deflate_lock);
+		wait_event(z_erofs_deflate_wq, READ_ONCE(z_erofs_deflate_head));
+		goto again;
+	}
+	z_erofs_deflate_head = strm->next;
+	spin_unlock(&z_erofs_deflate_lock);
+
+	/* 3. multi-call decompress */
+	zerr = zlib_inflateInit2(&strm->z, -MAX_WBITS);
+	if (zerr != Z_OK) {
+		reason = ERR_PTR(-EINVAL);
+		goto failed_zinit;
+	}
+
+	rq->fillgaps = true;	/* DEFLATE doesn't support NULL output buffer */
+	strm->z.avail_in = min(rq->inputsize, PAGE_SIZE - rq->pageofs_in);
+	rq->inputsize -= strm->z.avail_in;
+	strm->z.next_in = dctx.kin + rq->pageofs_in;
+	strm->z.avail_out = 0;
+	dctx.bounce = strm->bounce;
+
+	while (1) {
+		dctx.avail_out = strm->z.avail_out;
+		dctx.inbuf_sz = strm->z.avail_in;
+		reason = z_erofs_stream_switch_bufs(&dctx,
+					(void **)&strm->z.next_out,
+					(void **)&strm->z.next_in, pgpl);
+		if (reason)
+			break;
+		strm->z.avail_out = dctx.avail_out;
+		strm->z.avail_in = dctx.inbuf_sz;
+
+		zerr = zlib_inflate(&strm->z, Z_SYNC_FLUSH);
+		if (zerr != Z_OK || !(rq->outputsize + strm->z.avail_out)) {
+			if (zerr == Z_OK && rq->partial_decoding)
+				break;
+			if (zerr == Z_STREAM_END && !rq->outputsize)
+				break;
+			reason = (zerr == Z_DATA_ERROR ?
+				"corrupted compressed data" :
+				"unexpected end of stream");
+			break;
+		}
+	}
+	if (zlib_inflateEnd(&strm->z) != Z_OK && !reason)
+		reason = ERR_PTR(-EIO);
+	if (dctx.kout)
+		kunmap_local(dctx.kout);
+failed_zinit:
+	kunmap_local(dctx.kin);
+	/* 4. push back DEFLATE stream context to the global list */
+	spin_lock(&z_erofs_deflate_lock);
+	strm->next = z_erofs_deflate_head;
+	z_erofs_deflate_head = strm;
+	spin_unlock(&z_erofs_deflate_lock);
+	wake_up(&z_erofs_deflate_wq);
+	return reason;
+}
+
+static const char *z_erofs_deflate_decompress(struct z_erofs_decompress_req *rq,
+					      struct page **pgpl)
+{
+#ifdef CONFIG_EROFS_FS_ZIP_ACCEL
+	int err;
+
+	if (!rq->partial_decoding) {
+		err = z_erofs_crypto_decompress(rq, pgpl);
+		if (err != -EOPNOTSUPP)
+			return ERR_PTR(err);
+
+	}
+#endif
+	return __z_erofs_deflate_decompress(rq, pgpl);
+}
+
+const struct z_erofs_decompressor z_erofs_deflate_decomp = {
+	.config = z_erofs_load_deflate_config,
+	.decompress = z_erofs_deflate_decompress,
+	.init = z_erofs_deflate_init,
+	.exit = z_erofs_deflate_exit,
+	.name = "deflate",
+};
diff --git a/fs/erofs/decompressor_lzma.c b/fs/erofs/decompressor_lzma.c
new file mode 100644
index 000000000..6b0cdb446
--- /dev/null
+++ b/fs/erofs/decompressor_lzma.c
@@ -0,0 +1,236 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+#include <linux/xz.h>
+#include "compress.h"
+
+struct z_erofs_lzma {
+	struct z_erofs_lzma *next;
+	struct xz_dec_microlzma *state;
+	u8 bounce[PAGE_SIZE];
+};
+
+/* considering the LZMA performance, no need to use a lockless list for now */
+static DEFINE_SPINLOCK(z_erofs_lzma_lock);
+static unsigned int z_erofs_lzma_max_dictsize;
+static unsigned int z_erofs_lzma_nstrms, z_erofs_lzma_avail_strms;
+static struct z_erofs_lzma *z_erofs_lzma_head;
+static DECLARE_WAIT_QUEUE_HEAD(z_erofs_lzma_wq);
+
+module_param_named(lzma_streams, z_erofs_lzma_nstrms, uint, 0444);
+
+static void z_erofs_lzma_exit(void)
+{
+	/* there should be no running fs instance */
+	while (z_erofs_lzma_avail_strms) {
+		struct z_erofs_lzma *strm;
+
+		spin_lock(&z_erofs_lzma_lock);
+		strm = z_erofs_lzma_head;
+		if (!strm) {
+			spin_unlock(&z_erofs_lzma_lock);
+			DBG_BUGON(1);
+			return;
+		}
+		z_erofs_lzma_head = NULL;
+		spin_unlock(&z_erofs_lzma_lock);
+
+		while (strm) {
+			struct z_erofs_lzma *n = strm->next;
+
+			if (strm->state)
+				xz_dec_microlzma_end(strm->state);
+			kfree(strm);
+			--z_erofs_lzma_avail_strms;
+			strm = n;
+		}
+	}
+}
+
+static int __init z_erofs_lzma_init(void)
+{
+	unsigned int i;
+
+	/* by default, use # of possible CPUs instead */
+	if (!z_erofs_lzma_nstrms)
+		z_erofs_lzma_nstrms = min_t(unsigned int, num_possible_cpus(),
+				CONFIG_EROFS_FS_ZIP_LZMA_DEFAULT_MAX_STREAMS);
+
+	for (i = 0; i < z_erofs_lzma_nstrms; ++i) {
+		struct z_erofs_lzma *strm = kzalloc_obj(*strm);
+
+		if (!strm) {
+			z_erofs_lzma_exit();
+			return -ENOMEM;
+		}
+		spin_lock(&z_erofs_lzma_lock);
+		strm->next = z_erofs_lzma_head;
+		z_erofs_lzma_head = strm;
+		spin_unlock(&z_erofs_lzma_lock);
+		++z_erofs_lzma_avail_strms;
+	}
+	return 0;
+}
+
+static int z_erofs_load_lzma_config(struct super_block *sb,
+			struct erofs_super_block *dsb, void *data, int size)
+{
+	static DEFINE_MUTEX(lzma_resize_mutex);
+	struct z_erofs_lzma_cfgs *lzma = data;
+	unsigned int dict_size, i;
+	struct z_erofs_lzma *strm, *head = NULL;
+	int err;
+
+	if (!lzma || size < sizeof(struct z_erofs_lzma_cfgs)) {
+		erofs_err(sb, "invalid lzma cfgs, size=%u", size);
+		return -EINVAL;
+	}
+	if (lzma->format) {
+		erofs_err(sb, "unidentified lzma format %x, please check kernel version",
+			  le16_to_cpu(lzma->format));
+		return -EINVAL;
+	}
+	dict_size = le32_to_cpu(lzma->dict_size);
+	if (dict_size > Z_EROFS_LZMA_MAX_DICT_SIZE || dict_size < 4096) {
+		erofs_err(sb, "unsupported lzma dictionary size %u",
+			  dict_size);
+		return -EINVAL;
+	}
+
+	/* in case 2 z_erofs_load_lzma_config() race to avoid deadlock */
+	mutex_lock(&lzma_resize_mutex);
+
+	if (z_erofs_lzma_max_dictsize >= dict_size) {
+		mutex_unlock(&lzma_resize_mutex);
+		return 0;
+	}
+
+	/* 1. collect/isolate all streams for the following check */
+	for (i = 0; i < z_erofs_lzma_avail_strms; ++i) {
+		struct z_erofs_lzma *last;
+
+again:
+		spin_lock(&z_erofs_lzma_lock);
+		strm = z_erofs_lzma_head;
+		if (!strm) {
+			spin_unlock(&z_erofs_lzma_lock);
+			wait_event(z_erofs_lzma_wq,
+				   READ_ONCE(z_erofs_lzma_head));
+			goto again;
+		}
+		z_erofs_lzma_head = NULL;
+		spin_unlock(&z_erofs_lzma_lock);
+
+		for (last = strm; last->next; last = last->next)
+			++i;
+		last->next = head;
+		head = strm;
+	}
+
+	err = 0;
+	/* 2. walk each isolated stream and grow max dict_size if needed */
+	for (strm = head; strm; strm = strm->next) {
+		if (strm->state)
+			xz_dec_microlzma_end(strm->state);
+		strm->state = xz_dec_microlzma_alloc(XZ_PREALLOC, dict_size);
+		if (!strm->state)
+			err = -ENOMEM;
+	}
+
+	/* 3. push back all to the global list and update max dict_size */
+	spin_lock(&z_erofs_lzma_lock);
+	DBG_BUGON(z_erofs_lzma_head);
+	z_erofs_lzma_head = head;
+	spin_unlock(&z_erofs_lzma_lock);
+	wake_up_all(&z_erofs_lzma_wq);
+
+	z_erofs_lzma_max_dictsize = dict_size;
+	mutex_unlock(&lzma_resize_mutex);
+	return err;
+}
+
+static const char *z_erofs_lzma_decompress(struct z_erofs_decompress_req *rq,
+					   struct page **pgpl)
+{
+	struct super_block *sb = rq->sb;
+	struct z_erofs_stream_dctx dctx = { .rq = rq, .no = -1, .ni = 0 };
+	struct xz_buf buf = {};
+	struct z_erofs_lzma *strm;
+	enum xz_ret xz_err;
+	const char *reason;
+
+	/* 1. get the exact LZMA compressed size */
+	dctx.kin = kmap_local_page(*rq->in);
+	reason = z_erofs_fixup_insize(rq, dctx.kin + rq->pageofs_in,
+			min(rq->inputsize, sb->s_blocksize - rq->pageofs_in));
+	if (reason) {
+		kunmap_local(dctx.kin);
+		return reason;
+	}
+
+	/* 2. get an available lzma context */
+again:
+	spin_lock(&z_erofs_lzma_lock);
+	strm = z_erofs_lzma_head;
+	if (!strm) {
+		spin_unlock(&z_erofs_lzma_lock);
+		wait_event(z_erofs_lzma_wq, READ_ONCE(z_erofs_lzma_head));
+		goto again;
+	}
+	z_erofs_lzma_head = strm->next;
+	spin_unlock(&z_erofs_lzma_lock);
+
+	/* 3. multi-call decompress */
+	xz_dec_microlzma_reset(strm->state, rq->inputsize, rq->outputsize,
+			       !rq->partial_decoding);
+	buf.in_size = min(rq->inputsize, PAGE_SIZE - rq->pageofs_in);
+	rq->inputsize -= buf.in_size;
+	buf.in = dctx.kin + rq->pageofs_in;
+	dctx.bounce = strm->bounce;
+	do {
+		dctx.avail_out = buf.out_size - buf.out_pos;
+		dctx.inbuf_sz = buf.in_size;
+		dctx.inbuf_pos = buf.in_pos;
+		reason = z_erofs_stream_switch_bufs(&dctx, (void **)&buf.out,
+						    (void **)&buf.in, pgpl);
+		if (reason)
+			break;
+
+		if (buf.out_size == buf.out_pos) {
+			buf.out_size = dctx.avail_out;
+			buf.out_pos = 0;
+		}
+		buf.in_size = dctx.inbuf_sz;
+		buf.in_pos = dctx.inbuf_pos;
+
+		xz_err = xz_dec_microlzma_run(strm->state, &buf);
+		DBG_BUGON(buf.out_pos > buf.out_size);
+		DBG_BUGON(buf.in_pos > buf.in_size);
+
+		if (xz_err != XZ_OK) {
+			if (xz_err == XZ_STREAM_END && !rq->outputsize)
+				break;
+			reason = (xz_err == XZ_DATA_ERROR ?
+				"corrupted compressed data" :
+				"unexpected end of stream");
+			break;
+		}
+	} while (1);
+
+	if (dctx.kout)
+		kunmap_local(dctx.kout);
+	kunmap_local(dctx.kin);
+	/* 4. push back LZMA stream context to the global list */
+	spin_lock(&z_erofs_lzma_lock);
+	strm->next = z_erofs_lzma_head;
+	z_erofs_lzma_head = strm;
+	spin_unlock(&z_erofs_lzma_lock);
+	wake_up(&z_erofs_lzma_wq);
+	return reason;
+}
+
+const struct z_erofs_decompressor z_erofs_lzma_decomp = {
+	.config = z_erofs_load_lzma_config,
+	.decompress = z_erofs_lzma_decompress,
+	.init = z_erofs_lzma_init,
+	.exit = z_erofs_lzma_exit,
+	.name = "lzma"
+};
diff --git a/fs/erofs/decompressor_zstd.c b/fs/erofs/decompressor_zstd.c
new file mode 100644
index 000000000..ab318a240
--- /dev/null
+++ b/fs/erofs/decompressor_zstd.c
@@ -0,0 +1,220 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+#include <linux/zstd.h>
+#include "compress.h"
+
+struct z_erofs_zstd {
+	struct z_erofs_zstd *next;
+	u8 bounce[PAGE_SIZE];
+	void *wksp;
+	unsigned int wkspsz;
+};
+
+static DEFINE_SPINLOCK(z_erofs_zstd_lock);
+static unsigned int z_erofs_zstd_max_dictsize;
+static unsigned int z_erofs_zstd_nstrms, z_erofs_zstd_avail_strms;
+static struct z_erofs_zstd *z_erofs_zstd_head;
+static DECLARE_WAIT_QUEUE_HEAD(z_erofs_zstd_wq);
+
+module_param_named(zstd_streams, z_erofs_zstd_nstrms, uint, 0444);
+
+static struct z_erofs_zstd *z_erofs_isolate_strms(bool all)
+{
+	struct z_erofs_zstd *strm;
+
+again:
+	spin_lock(&z_erofs_zstd_lock);
+	strm = z_erofs_zstd_head;
+	if (!strm) {
+		spin_unlock(&z_erofs_zstd_lock);
+		wait_event(z_erofs_zstd_wq, READ_ONCE(z_erofs_zstd_head));
+		goto again;
+	}
+	z_erofs_zstd_head = all ? NULL : strm->next;
+	spin_unlock(&z_erofs_zstd_lock);
+	return strm;
+}
+
+static void z_erofs_zstd_exit(void)
+{
+	while (z_erofs_zstd_avail_strms) {
+		struct z_erofs_zstd *strm, *n;
+
+		for (strm = z_erofs_isolate_strms(true); strm; strm = n) {
+			n = strm->next;
+
+			kvfree(strm->wksp);
+			kfree(strm);
+			--z_erofs_zstd_avail_strms;
+		}
+	}
+}
+
+static int __init z_erofs_zstd_init(void)
+{
+	/* by default, use # of possible CPUs instead */
+	if (!z_erofs_zstd_nstrms)
+		z_erofs_zstd_nstrms = num_possible_cpus();
+
+	for (; z_erofs_zstd_avail_strms < z_erofs_zstd_nstrms;
+	     ++z_erofs_zstd_avail_strms) {
+		struct z_erofs_zstd *strm;
+
+		strm = kzalloc_obj(*strm);
+		if (!strm) {
+			z_erofs_zstd_exit();
+			return -ENOMEM;
+		}
+		spin_lock(&z_erofs_zstd_lock);
+		strm->next = z_erofs_zstd_head;
+		z_erofs_zstd_head = strm;
+		spin_unlock(&z_erofs_zstd_lock);
+	}
+	return 0;
+}
+
+static int z_erofs_load_zstd_config(struct super_block *sb,
+			struct erofs_super_block *dsb, void *data, int size)
+{
+	static DEFINE_MUTEX(zstd_resize_mutex);
+	struct z_erofs_zstd_cfgs *zstd = data;
+	unsigned int dict_size, wkspsz;
+	struct z_erofs_zstd *strm, *head = NULL;
+	void *wksp;
+
+	if (!zstd || size < sizeof(struct z_erofs_zstd_cfgs) || zstd->format) {
+		erofs_err(sb, "unsupported zstd format, size=%u", size);
+		return -EINVAL;
+	}
+
+	if (zstd->windowlog > ilog2(Z_EROFS_ZSTD_MAX_DICT_SIZE) - 10) {
+		erofs_err(sb, "unsupported zstd window log %u", zstd->windowlog);
+		return -EINVAL;
+	}
+	dict_size = 1U << (zstd->windowlog + 10);
+
+	/* in case 2 z_erofs_load_zstd_config() race to avoid deadlock */
+	mutex_lock(&zstd_resize_mutex);
+	if (z_erofs_zstd_max_dictsize >= dict_size) {
+		mutex_unlock(&zstd_resize_mutex);
+		return 0;
+	}
+
+	/* 1. collect/isolate all streams for the following check */
+	while (z_erofs_zstd_avail_strms) {
+		struct z_erofs_zstd *n;
+
+		for (strm = z_erofs_isolate_strms(true); strm; strm = n) {
+			n = strm->next;
+			strm->next = head;
+			head = strm;
+			--z_erofs_zstd_avail_strms;
+		}
+	}
+
+	/* 2. walk each isolated stream and grow max dict_size if needed */
+	wkspsz = zstd_dstream_workspace_bound(dict_size);
+	for (strm = head; strm; strm = strm->next) {
+		wksp = kvmalloc(wkspsz, GFP_KERNEL);
+		if (!wksp)
+			break;
+		kvfree(strm->wksp);
+		strm->wksp = wksp;
+		strm->wkspsz = wkspsz;
+	}
+
+	/* 3. push back all to the global list and update max dict_size */
+	spin_lock(&z_erofs_zstd_lock);
+	DBG_BUGON(z_erofs_zstd_head);
+	z_erofs_zstd_head = head;
+	spin_unlock(&z_erofs_zstd_lock);
+	z_erofs_zstd_avail_strms = z_erofs_zstd_nstrms;
+	wake_up_all(&z_erofs_zstd_wq);
+	if (!strm)
+		z_erofs_zstd_max_dictsize = dict_size;
+	mutex_unlock(&zstd_resize_mutex);
+	return strm ? -ENOMEM : 0;
+}
+
+static const char *z_erofs_zstd_decompress(struct z_erofs_decompress_req *rq,
+					   struct page **pgpl)
+{
+	struct super_block *sb = rq->sb;
+	struct z_erofs_stream_dctx dctx = { .rq = rq, .no = -1, .ni = 0 };
+	zstd_in_buffer in_buf = { NULL, 0, 0 };
+	zstd_out_buffer out_buf = { NULL, 0, 0 };
+	struct z_erofs_zstd *strm;
+	zstd_dstream *stream;
+	const char *reason;
+	int zerr;
+
+	/* 1. get the exact compressed size */
+	dctx.kin = kmap_local_page(*rq->in);
+	reason = z_erofs_fixup_insize(rq, dctx.kin + rq->pageofs_in,
+			min(rq->inputsize, sb->s_blocksize - rq->pageofs_in));
+	if (reason) {
+		kunmap_local(dctx.kin);
+		return reason;
+	}
+
+	/* 2. get an available ZSTD context */
+	strm = z_erofs_isolate_strms(false);
+
+	/* 3. multi-call decompress */
+	stream = zstd_init_dstream(z_erofs_zstd_max_dictsize, strm->wksp, strm->wkspsz);
+	if (!stream) {
+		reason = ERR_PTR(-ENOMEM);
+		goto failed_zinit;
+	}
+
+	rq->fillgaps = true;	/* ZSTD doesn't support NULL output buffer */
+	in_buf.size = min_t(u32, rq->inputsize, PAGE_SIZE - rq->pageofs_in);
+	rq->inputsize -= in_buf.size;
+	in_buf.src = dctx.kin + rq->pageofs_in;
+	dctx.bounce = strm->bounce;
+
+	do {
+		dctx.inbuf_sz = in_buf.size;
+		dctx.inbuf_pos = in_buf.pos;
+		reason = z_erofs_stream_switch_bufs(&dctx, &out_buf.dst,
+						 (void **)&in_buf.src, pgpl);
+		if (reason)
+			break;
+
+		if (out_buf.size == out_buf.pos) {
+			out_buf.size = dctx.avail_out;
+			out_buf.pos = 0;
+		}
+		in_buf.size = dctx.inbuf_sz;
+		in_buf.pos = dctx.inbuf_pos;
+
+		zerr = zstd_decompress_stream(stream, &out_buf, &in_buf);
+		dctx.avail_out = out_buf.size - out_buf.pos;
+		if (zstd_is_error(zerr) ||
+		    ((rq->outputsize + dctx.avail_out) && (!zerr || (zerr > 0 &&
+				!(rq->inputsize + in_buf.size - in_buf.pos))))) {
+			reason = zstd_is_error(zerr) ? zstd_get_error_name(zerr) :
+					"unexpected end of stream";
+			break;
+		}
+	} while (rq->outputsize + dctx.avail_out);
+
+	if (dctx.kout)
+		kunmap_local(dctx.kout);
+failed_zinit:
+	kunmap_local(dctx.kin);
+	/* 4. push back ZSTD stream context to the global list */
+	spin_lock(&z_erofs_zstd_lock);
+	strm->next = z_erofs_zstd_head;
+	z_erofs_zstd_head = strm;
+	spin_unlock(&z_erofs_zstd_lock);
+	wake_up(&z_erofs_zstd_wq);
+	return reason;
+}
+
+const struct z_erofs_decompressor z_erofs_zstd_decomp = {
+	.config = z_erofs_load_zstd_config,
+	.decompress = z_erofs_zstd_decompress,
+	.init = z_erofs_zstd_init,
+	.exit = z_erofs_zstd_exit,
+	.name = "zstd",
+};
diff --git a/fs/erofs/dir.c b/fs/erofs/dir.c
new file mode 100644
index 000000000..4aa52a5f2
--- /dev/null
+++ b/fs/erofs/dir.c
@@ -0,0 +1,134 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2017-2018 HUAWEI, Inc.
+ *             https://www.huawei.com/
+ * Copyright (C) 2022, Alibaba Cloud
+ */
+#include "internal.h"
+#include <linux/filelock.h>
+
+static int erofs_fill_dentries(struct inode *dir, struct dir_context *ctx,
+			       void *dentry_blk, struct erofs_dirent *de,
+			       unsigned int nameoff0, unsigned int maxsize)
+{
+	const struct erofs_dirent *end = dentry_blk + nameoff0;
+
+	while (de < end) {
+		unsigned char d_type = fs_ftype_to_dtype(de->file_type);
+		unsigned int nameoff = le16_to_cpu(de->nameoff);
+		const char *de_name = (char *)dentry_blk + nameoff;
+		unsigned int de_namelen;
+
+		/* non-trailing dirent in the directory block? */
+		if (de + 1 < end)
+			de_namelen = le16_to_cpu(de[1].nameoff) - nameoff;
+		else if (maxsize <= nameoff)
+			goto err_bogus;
+		else
+			de_namelen = strnlen(de_name, maxsize - nameoff);
+
+		/* a corrupted entry is found (including negative namelen) */
+		if (!in_range32(de_namelen, 1, EROFS_NAME_LEN) ||
+		    nameoff + de_namelen > maxsize)
+			goto err_bogus;
+
+		if (!dir_emit(ctx, de_name, de_namelen,
+			      erofs_nid_to_ino64(EROFS_SB(dir->i_sb),
+						 le64_to_cpu(de->nid)), d_type))
+			return 1;
+		++de;
+		ctx->pos += sizeof(struct erofs_dirent);
+	}
+	return 0;
+err_bogus:
+	erofs_err(dir->i_sb, "bogus dirent @ nid %llu", EROFS_I(dir)->nid);
+	DBG_BUGON(1);
+	return -EFSCORRUPTED;
+}
+
+static int erofs_readdir(struct file *f, struct dir_context *ctx)
+{
+	struct inode *dir = file_inode(f);
+	struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
+	struct super_block *sb = dir->i_sb;
+	struct file_ra_state *ra = &f->f_ra;
+	unsigned long bsz = sb->s_blocksize;
+	unsigned int ofs = erofs_blkoff(sb, ctx->pos);
+	pgoff_t ra_pages = DIV_ROUND_UP_POW2(
+			EROFS_I_SB(dir)->dir_ra_bytes, PAGE_SIZE);
+	pgoff_t nr_pages = DIV_ROUND_UP_POW2(dir->i_size, PAGE_SIZE);
+	int err = 0;
+	bool initial = true;
+
+	buf.mapping = dir->i_mapping;
+	while (ctx->pos < dir->i_size) {
+		erofs_off_t dbstart = ctx->pos - ofs;
+		struct erofs_dirent *de;
+		unsigned int nameoff, maxsize;
+
+		if (fatal_signal_pending(current)) {
+			err = -ERESTARTSYS;
+			break;
+		}
+
+		/* readahead blocks to enhance performance for large directories */
+		if (ra_pages) {
+			pgoff_t idx = DIV_ROUND_UP_POW2(ctx->pos, PAGE_SIZE);
+			pgoff_t pages = min(nr_pages - idx, ra_pages);
+
+			if (pages > 1 && !ra_has_index(ra, idx))
+				page_cache_sync_readahead(dir->i_mapping, ra,
+							  f, idx, pages);
+		}
+
+		de = erofs_bread(&buf, dbstart, true);
+		if (IS_ERR(de)) {
+			erofs_err(sb, "failed to readdir of logical block %llu of nid %llu",
+				  erofs_blknr(sb, dbstart), EROFS_I(dir)->nid);
+			err = PTR_ERR(de);
+			break;
+		}
+
+		nameoff = le16_to_cpu(de->nameoff);
+		if (!nameoff || nameoff >= bsz || (nameoff % sizeof(*de))) {
+			erofs_err(sb, "invalid de[0].nameoff %u @ nid %llu",
+				  nameoff, EROFS_I(dir)->nid);
+			err = -EFSCORRUPTED;
+			break;
+		}
+
+		maxsize = min_t(unsigned int, dir->i_size - dbstart, bsz);
+		/* search dirents at the arbitrary position */
+		if (initial) {
+			initial = false;
+			ofs = roundup(ofs, sizeof(struct erofs_dirent));
+			ctx->pos = dbstart + ofs;
+		}
+
+		err = erofs_fill_dentries(dir, ctx, de, (void *)de + ofs,
+					  nameoff, maxsize);
+		if (err)
+			break;
+		ctx->pos = dbstart + maxsize;
+		ofs = 0;
+		cond_resched();
+	}
+	erofs_put_metabuf(&buf);
+	if (EROFS_I(dir)->dot_omitted && ctx->pos == dir->i_size) {
+		if (!dir_emit_dot(f, ctx))
+			return 0;
+		++ctx->pos;
+	}
+	return err < 0 ? err : 0;
+}
+
+const struct file_operations erofs_dir_fops = {
+	.llseek		= generic_file_llseek,
+	.read		= generic_read_dir,
+	.iterate_shared	= erofs_readdir,
+	.unlocked_ioctl = erofs_ioctl,
+#ifdef CONFIG_COMPAT
+	.compat_ioctl   = erofs_compat_ioctl,
+#endif
+	.setlease	= generic_setlease,
+};
diff --git a/fs/erofs/erofs_fs.h b/fs/erofs/erofs_fs.h
new file mode 100644
index 000000000..16ec4fd33
--- /dev/null
+++ b/fs/erofs/erofs_fs.h
@@ -0,0 +1,469 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * EROFS (Enhanced ROM File System) on-disk format definition
+ *
+ * Copyright (C) 2017-2018 HUAWEI, Inc.
+ *             https://www.huawei.com/
+ * Copyright (C) 2021, Alibaba Cloud
+ */
+#ifndef __EROFS_FS_H
+#define __EROFS_FS_H
+
+/* to allow for x86 boot sectors and other oddities. */
+#define EROFS_SUPER_OFFSET      1024
+
+#define EROFS_FEATURE_COMPAT_SB_CHKSUM			0x00000001
+#define EROFS_FEATURE_COMPAT_MTIME			0x00000002
+#define EROFS_FEATURE_COMPAT_XATTR_FILTER		0x00000004
+#define EROFS_FEATURE_COMPAT_SHARED_EA_IN_METABOX	0x00000008
+#define EROFS_FEATURE_COMPAT_PLAIN_XATTR_PFX		0x00000010
+#define EROFS_FEATURE_COMPAT_ISHARE_XATTRS		0x00000020
+
+/*
+ * Any bits that aren't in EROFS_ALL_FEATURE_INCOMPAT should
+ * be incompatible with this kernel version.
+ */
+#define EROFS_FEATURE_INCOMPAT_LZ4_0PADDING	0x00000001
+#define EROFS_FEATURE_INCOMPAT_COMPR_CFGS	0x00000002
+#define EROFS_FEATURE_INCOMPAT_BIG_PCLUSTER	0x00000002
+#define EROFS_FEATURE_INCOMPAT_CHUNKED_FILE	0x00000004
+#define EROFS_FEATURE_INCOMPAT_DEVICE_TABLE	0x00000008
+#define EROFS_FEATURE_INCOMPAT_COMPR_HEAD2	0x00000008
+#define EROFS_FEATURE_INCOMPAT_ZTAILPACKING	0x00000010
+#define EROFS_FEATURE_INCOMPAT_FRAGMENTS	0x00000020
+#define EROFS_FEATURE_INCOMPAT_DEDUPE		0x00000020
+#define EROFS_FEATURE_INCOMPAT_XATTR_PREFIXES	0x00000040
+#define EROFS_FEATURE_INCOMPAT_48BIT		0x00000080
+#define EROFS_FEATURE_INCOMPAT_METABOX		0x00000100
+#define EROFS_ALL_FEATURE_INCOMPAT		\
+	((EROFS_FEATURE_INCOMPAT_METABOX << 1) - 1)
+
+#define EROFS_SB_EXTSLOT_SIZE	16
+
+struct erofs_deviceslot {
+	u8 tag[64];		/* digest(sha256), etc. */
+	__le32 blocks_lo;	/* total blocks count of this device */
+	__le32 uniaddr_lo;	/* unified starting block of this device */
+	__le16 blocks_hi;	/* total blocks count MSB */
+	__le16 uniaddr_hi;	/* unified starting block MSB */
+	u8 reserved[52];
+};
+#define EROFS_DEVT_SLOT_SIZE	sizeof(struct erofs_deviceslot)
+
+/* erofs on-disk super block (currently 144 bytes at maximum) */
+struct erofs_super_block {
+	__le32 magic;           /* file system magic number */
+	__le32 checksum;        /* crc32c to avoid unexpected on-disk overlap */
+	__le32 feature_compat;
+	__u8 blkszbits;         /* filesystem block size in bit shift */
+	__u8 sb_extslots;	/* superblock size = 128 + sb_extslots * 16 */
+	union {
+		__le16 rootnid_2b;	/* nid of root directory */
+		__le16 blocks_hi;	/* (48BIT on) blocks count MSB */
+	} __packed rb;
+	__le64 inos;            /* total valid ino # (== f_files - f_favail) */
+	__le64 epoch;		/* base seconds used for compact inodes */
+	__le32 fixed_nsec;	/* fixed nanoseconds for compact inodes */
+	__le32 blocks_lo;	/* blocks count LSB */
+	__le32 meta_blkaddr;	/* start block address of metadata area */
+	__le32 xattr_blkaddr;	/* start block address of shared xattr area */
+	__u8 uuid[16];          /* 128-bit uuid for volume */
+	__u8 volume_name[16];   /* volume name */
+	__le32 feature_incompat;
+	union {
+		/* bitmap for available compression algorithms */
+		__le16 available_compr_algs;
+		/* customized sliding window size instead of 64k by default */
+		__le16 lz4_max_distance;
+	} __packed u1;
+	__le16 extra_devices;	/* # of devices besides the primary device */
+	__le16 devt_slotoff;	/* startoff = devt_slotoff * devt_slotsize */
+	__u8 dirblkbits;	/* directory block size in bit shift */
+	__u8 xattr_prefix_count;	/* # of long xattr name prefixes */
+	__le32 xattr_prefix_start;	/* start of long xattr prefixes */
+	__le64 packed_nid;	/* nid of the special packed inode */
+	__u8 xattr_filter_reserved; /* reserved for xattr name filter */
+	__u8 ishare_xattr_prefix_id;
+	__u8 reserved[2];
+	__le32 build_time;	/* seconds added to epoch for mkfs time */
+	__le64 rootnid_8b;	/* (48BIT on) nid of root directory */
+	__le64 reserved2;
+	__le64 metabox_nid;     /* (METABOX on) nid of the metabox inode */
+	__le64 reserved3;	/* [align to extslot 1] */
+};
+
+/*
+ * EROFS inode datalayout (i_format in on-disk inode):
+ * 0 - uncompressed flat inode without tail-packing inline data:
+ * 1 - compressed inode with non-compact indexes:
+ * 2 - uncompressed flat inode with tail-packing inline data:
+ * 3 - compressed inode with compact indexes:
+ * 4 - chunk-based inode with (optional) multi-device support:
+ * 5~7 - reserved
+ */
+enum {
+	EROFS_INODE_FLAT_PLAIN			= 0,
+	EROFS_INODE_COMPRESSED_FULL		= 1,
+	EROFS_INODE_FLAT_INLINE			= 2,
+	EROFS_INODE_COMPRESSED_COMPACT		= 3,
+	EROFS_INODE_CHUNK_BASED			= 4,
+	EROFS_INODE_DATALAYOUT_MAX
+};
+
+static inline bool erofs_inode_is_data_compressed(unsigned int datamode)
+{
+	return datamode == EROFS_INODE_COMPRESSED_COMPACT ||
+		datamode == EROFS_INODE_COMPRESSED_FULL;
+}
+
+/* bit definitions of inode i_format */
+#define EROFS_I_VERSION_MASK            0x01
+#define EROFS_I_DATALAYOUT_MASK         0x07
+
+#define EROFS_I_VERSION_BIT	0
+#define EROFS_I_DATALAYOUT_BIT	1
+#define EROFS_I_NLINK_1_BIT	4	/* non-directory compact inodes only */
+#define EROFS_I_DOT_OMITTED_BIT	4	/* (directories) omit the `.` dirent */
+#define EROFS_I_ALL		((1 << (EROFS_I_NLINK_1_BIT + 1)) - 1)
+
+/* indicate chunk blkbits, thus 'chunksize = blocksize << chunk blkbits' */
+#define EROFS_CHUNK_FORMAT_BLKBITS_MASK		0x001F
+/* with chunk indexes or just a 4-byte block array */
+#define EROFS_CHUNK_FORMAT_INDEXES		0x0020
+#define EROFS_CHUNK_FORMAT_48BIT		0x0040
+
+#define EROFS_CHUNK_FORMAT_ALL	((EROFS_CHUNK_FORMAT_48BIT << 1) - 1)
+
+/* 32-byte on-disk inode */
+#define EROFS_INODE_LAYOUT_COMPACT	0
+/* 64-byte on-disk inode */
+#define EROFS_INODE_LAYOUT_EXTENDED	1
+
+struct erofs_inode_chunk_info {
+	__le16 format;		/* chunk blkbits, etc. */
+	__le16 reserved;
+};
+
+union erofs_inode_i_u {
+	__le32 blocks_lo;	/* total blocks count (if compressed inodes) */
+	__le32 startblk_lo;	/* starting block number (if flat inodes) */
+	__le32 rdev;		/* device ID (if special inodes) */
+	struct erofs_inode_chunk_info c;
+};
+
+union erofs_inode_i_nb {
+	__le16 nlink;		/* if EROFS_I_NLINK_1_BIT is unset */
+	__le16 blocks_hi;	/* total blocks count MSB */
+	__le16 startblk_hi;	/* starting block number MSB */
+} __packed;
+
+/* 32-byte reduced form of an ondisk inode */
+struct erofs_inode_compact {
+	__le16 i_format;	/* inode format hints */
+	__le16 i_xattr_icount;
+	__le16 i_mode;
+	union erofs_inode_i_nb i_nb;
+	__le32 i_size;
+	__le32 i_mtime;
+	union erofs_inode_i_u i_u;
+
+	__le32 i_ino;		/* only used for 32-bit stat compatibility */
+	__le16 i_uid;
+	__le16 i_gid;
+	__le32 i_reserved;
+};
+
+/* 64-byte complete form of an ondisk inode */
+struct erofs_inode_extended {
+	__le16 i_format;	/* inode format hints */
+	__le16 i_xattr_icount;
+	__le16 i_mode;
+	union erofs_inode_i_nb i_nb;
+	__le64 i_size;
+	union erofs_inode_i_u i_u;
+
+	__le32 i_ino;		/* only used for 32-bit stat compatibility */
+	__le32 i_uid;
+	__le32 i_gid;
+	__le64 i_mtime;
+	__le32 i_mtime_nsec;
+	__le32 i_nlink;
+	__u8   i_reserved2[16];
+};
+
+/*
+ * inline xattrs (n == i_xattr_icount):
+ * erofs_xattr_ibody_header(1) + (n - 1) * 4 bytes
+ *          12 bytes           /                   \
+ *                            /                     \
+ *                           /-----------------------\
+ *                           |  erofs_xattr_entries+ |
+ *                           +-----------------------+
+ * inline xattrs must starts in erofs_xattr_ibody_header,
+ * for read-only fs, no need to introduce h_refcount
+ */
+struct erofs_xattr_ibody_header {
+	__le32 h_name_filter;		/* bit value 1 indicates not-present */
+	__u8   h_shared_count;
+	__u8   h_reserved2[7];
+	__le32 h_shared_xattrs[];       /* shared xattr id array */
+};
+
+/* Name indexes */
+#define EROFS_XATTR_INDEX_USER              1
+#define EROFS_XATTR_INDEX_POSIX_ACL_ACCESS  2
+#define EROFS_XATTR_INDEX_POSIX_ACL_DEFAULT 3
+#define EROFS_XATTR_INDEX_TRUSTED           4
+#define EROFS_XATTR_INDEX_LUSTRE            5
+#define EROFS_XATTR_INDEX_SECURITY          6
+
+/*
+ * bit 7 of e_name_index is set when it refers to a long xattr name prefix,
+ * while the remained lower bits represent the index of the prefix.
+ */
+#define EROFS_XATTR_LONG_PREFIX		0x80
+#define EROFS_XATTR_LONG_PREFIX_MASK	0x7f
+
+#define EROFS_XATTR_FILTER_BITS		32
+#define EROFS_XATTR_FILTER_DEFAULT	UINT32_MAX
+#define EROFS_XATTR_FILTER_SEED		0x25BBE08F
+
+/* xattr entry (for both inline & shared xattrs) */
+struct erofs_xattr_entry {
+	__u8   e_name_len;      /* length of name */
+	__u8   e_name_index;    /* attribute name index */
+	__le16 e_value_size;    /* size of attribute value */
+	/* followed by e_name and e_value */
+	char   e_name[];        /* attribute name */
+};
+
+/* long xattr name prefix */
+struct erofs_xattr_long_prefix {
+	__u8   base_index;	/* short xattr name prefix index */
+	char   infix[];		/* infix apart from short prefix */
+};
+
+static inline unsigned int erofs_xattr_ibody_size(__le16 i_xattr_icount)
+{
+	if (!i_xattr_icount)
+		return 0;
+
+	/* 1 header + n-1 * 4 bytes inline xattr to keep continuity */
+	return sizeof(struct erofs_xattr_ibody_header) +
+		sizeof(__u32) * (le16_to_cpu(i_xattr_icount) - 1);
+}
+
+#define EROFS_XATTR_ALIGN(size) round_up(size, sizeof(struct erofs_xattr_entry))
+
+static inline unsigned int erofs_xattr_entry_size(struct erofs_xattr_entry *e)
+{
+	return EROFS_XATTR_ALIGN(sizeof(struct erofs_xattr_entry) +
+				 e->e_name_len + le16_to_cpu(e->e_value_size));
+}
+
+/* represent a zeroed chunk (hole) */
+#define EROFS_NULL_ADDR			-1
+
+/* 4-byte block address array */
+#define EROFS_BLOCK_MAP_ENTRY_SIZE	sizeof(__le32)
+
+/* 8-byte inode chunk index */
+struct erofs_inode_chunk_index {
+	__le16 startblk_hi;	/* starting block number MSB */
+	__le16 device_id;	/* back-end storage id (with bits masked) */
+	__le32 startblk_lo;	/* starting block number of this chunk */
+};
+
+#define EROFS_DIRENT_NID_METABOX_BIT	63
+#define EROFS_DIRENT_NID_MASK	(BIT_ULL(EROFS_DIRENT_NID_METABOX_BIT) - 1)
+
+/* dirent sorts in alphabet order, thus we can do binary search */
+struct erofs_dirent {
+	__le64 nid;     /* node number */
+	__le16 nameoff; /* start offset of file name */
+	__u8 file_type; /* file type */
+	__u8 reserved;  /* reserved */
+} __packed;
+
+/*
+ * EROFS file types should match generic FT_* types and
+ * it seems no need to add BUILD_BUG_ONs since potential
+ * unmatchness will break other fses as well...
+ */
+
+#define EROFS_NAME_LEN      255
+
+/* maximum supported encoded size of a physical compressed cluster */
+#define Z_EROFS_PCLUSTER_MAX_SIZE	(1024 * 1024)
+
+/* maximum supported decoded size of a physical compressed cluster */
+#define Z_EROFS_PCLUSTER_MAX_DSIZE	(12 * 1024 * 1024)
+
+/* available compression algorithm types (for h_algorithmtype) */
+enum {
+	Z_EROFS_COMPRESSION_LZ4		= 0,
+	Z_EROFS_COMPRESSION_LZMA	= 1,
+	Z_EROFS_COMPRESSION_DEFLATE	= 2,
+	Z_EROFS_COMPRESSION_ZSTD	= 3,
+	Z_EROFS_COMPRESSION_MAX
+};
+#define Z_EROFS_ALL_COMPR_ALGS		((1 << Z_EROFS_COMPRESSION_MAX) - 1)
+
+/* 14 bytes (+ length field = 16 bytes) */
+struct z_erofs_lz4_cfgs {
+	__le16 max_distance;
+	__le16 max_pclusterblks;
+	u8 reserved[10];
+} __packed;
+
+/* 14 bytes (+ length field = 16 bytes) */
+struct z_erofs_lzma_cfgs {
+	__le32 dict_size;
+	__le16 format;
+	u8 reserved[8];
+} __packed;
+
+#define Z_EROFS_LZMA_MAX_DICT_SIZE	(8 * Z_EROFS_PCLUSTER_MAX_SIZE)
+
+/* 6 bytes (+ length field = 8 bytes) */
+struct z_erofs_deflate_cfgs {
+	u8 windowbits;			/* 8..15 for DEFLATE */
+	u8 reserved[5];
+} __packed;
+
+/* 6 bytes (+ length field = 8 bytes) */
+struct z_erofs_zstd_cfgs {
+	u8 format;
+	u8 windowlog;           /* windowLog - ZSTD_WINDOWLOG_ABSOLUTEMIN(10) */
+	u8 reserved[4];
+} __packed;
+
+#define Z_EROFS_ZSTD_MAX_DICT_SIZE      Z_EROFS_PCLUSTER_MAX_SIZE
+
+/*
+ * Enable COMPACTED_2B for EROFS_INODE_COMPRESSED_COMPACT inodes:
+ *   4B (disabled) vs 4B+2B+4B (enabled)
+ */
+#define Z_EROFS_ADVISE_COMPACTED_2B		0x0001
+/* Enable extent metadata for EROFS_INODE_COMPRESSED_FULL inodes */
+#define Z_EROFS_ADVISE_EXTENTS			0x0001
+#define Z_EROFS_ADVISE_BIG_PCLUSTER_1		0x0002
+#define Z_EROFS_ADVISE_BIG_PCLUSTER_2		0x0004
+#define Z_EROFS_ADVISE_INLINE_PCLUSTER		0x0008
+#define Z_EROFS_ADVISE_INTERLACED_PCLUSTER	0x0010
+#define Z_EROFS_ADVISE_FRAGMENT_PCLUSTER	0x0020
+/* Indicate the record size for each extent if extent metadata is used */
+#define Z_EROFS_ADVISE_EXTRECSZ_BIT		1
+#define Z_EROFS_ADVISE_EXTRECSZ_MASK		0x3
+
+#define Z_EROFS_FRAGMENT_INODE_BIT              7
+struct z_erofs_map_header {
+	union {
+		/* fragment data offset in the packed inode */
+		__le32  h_fragmentoff;
+		struct {
+			__le16  h_reserved1;
+			/* indicates the encoded size of tailpacking data */
+			__le16  h_idata_size;
+		};
+		__le32 h_extents_lo;	/* extent count LSB */
+	};
+	__le16	h_advise;
+	union {
+		struct {
+			/* algorithm type (bit 0-3: HEAD1; bit 4-7: HEAD2) */
+			__u8	h_algorithmtype;
+			/*
+			 * bit 0-3 : logical cluster bits - blkszbits
+			 * bit 4-6 : reserved
+			 * bit 7   : pack the whole file into packed inode
+			 */
+			__u8	h_clusterbits;
+		} __packed;
+		__le16 h_extents_hi;	/* extent count MSB */
+	} __packed;
+};
+
+enum {
+	Z_EROFS_LCLUSTER_TYPE_PLAIN	= 0,
+	Z_EROFS_LCLUSTER_TYPE_HEAD1	= 1,
+	Z_EROFS_LCLUSTER_TYPE_NONHEAD	= 2,
+	Z_EROFS_LCLUSTER_TYPE_HEAD2	= 3,
+	Z_EROFS_LCLUSTER_TYPE_MAX
+};
+
+#define Z_EROFS_LI_LCLUSTER_TYPE_MASK	(Z_EROFS_LCLUSTER_TYPE_MAX - 1)
+
+/* (noncompact only, HEAD) This pcluster refers to partial decompressed data */
+#define Z_EROFS_LI_PARTIAL_REF		(1 << 15)
+/* (noncompact only, HEAD) This pcluster can also be regarded as a HOLE */
+#define Z_EROFS_LI_HOLE			(1 << 14)
+
+/* Set on 1st non-head lcluster to store compressed block counti (in blocks) */
+#define Z_EROFS_LI_D0_CBLKCNT		(1 << 11)
+
+struct z_erofs_lcluster_index {
+	__le16 di_advise;
+	/* where to decompress in the head lcluster */
+	__le16 di_clusterofs;
+
+	union {
+		__le32 blkaddr;		/* for the HEAD lclusters */
+		/*
+		 * [0] - distance to its HEAD lcluster
+		 * [1] - distance to the next HEAD lcluster
+		 */
+		__le16 delta[2];	/* for the NONHEAD lclusters */
+	} di_u;
+};
+
+#define Z_EROFS_MAP_HEADER_END(end)	\
+	(ALIGN(end, 8) + sizeof(struct z_erofs_map_header))
+#define Z_EROFS_FULL_INDEX_START(end)	(Z_EROFS_MAP_HEADER_END(end) + 8)
+
+#define Z_EROFS_EXTENT_PLEN_PARTIAL	BIT(27)
+#define Z_EROFS_EXTENT_PLEN_FMT_BIT	28
+#define Z_EROFS_EXTENT_PLEN_MASK	((Z_EROFS_PCLUSTER_MAX_SIZE << 1) - 1)
+struct z_erofs_extent {
+	__le32 plen;		/* encoded length */
+	__le32 pstart_lo;	/* physical offset */
+	__le32 pstart_hi;	/* physical offset MSB */
+	__le32 lstart_lo;	/* logical offset */
+	__le32 lstart_hi;	/* logical offset MSB (>= 4GiB inodes) */
+	__u8 reserved[12];	/* for future use */
+};
+
+static inline int z_erofs_extent_recsize(unsigned int advise)
+{
+	return 4 << ((advise >> Z_EROFS_ADVISE_EXTRECSZ_BIT) &
+		Z_EROFS_ADVISE_EXTRECSZ_MASK);
+}
+
+/* check the EROFS on-disk layout strictly at compile time */
+static inline void erofs_check_ondisk_layout_definitions(void)
+{
+	const __le64 fmh = *(__le64 *)&(struct z_erofs_map_header) {
+		.h_clusterbits = 1 << Z_EROFS_FRAGMENT_INODE_BIT
+	};
+
+	BUILD_BUG_ON(sizeof(struct erofs_super_block) != 144);
+	BUILD_BUG_ON(sizeof(struct erofs_inode_compact) != 32);
+	BUILD_BUG_ON(sizeof(struct erofs_inode_extended) != 64);
+	BUILD_BUG_ON(sizeof(struct erofs_xattr_ibody_header) != 12);
+	BUILD_BUG_ON(sizeof(struct erofs_xattr_entry) != 4);
+	BUILD_BUG_ON(sizeof(struct erofs_inode_chunk_info) != 4);
+	BUILD_BUG_ON(sizeof(struct erofs_inode_chunk_index) != 8);
+	BUILD_BUG_ON(sizeof(struct z_erofs_map_header) != 8);
+	BUILD_BUG_ON(sizeof(struct z_erofs_lcluster_index) != 8);
+	BUILD_BUG_ON(sizeof(struct erofs_dirent) != 12);
+	/* keep in sync between 2 index structures for better extendibility */
+	BUILD_BUG_ON(sizeof(struct erofs_inode_chunk_index) !=
+		     sizeof(struct z_erofs_lcluster_index));
+	BUILD_BUG_ON(sizeof(struct erofs_deviceslot) != 128);
+
+	/* exclude old compiler versions like gcc 7.5.0 */
+	BUILD_BUG_ON(__builtin_constant_p(fmh) ?
+		     fmh != cpu_to_le64(1ULL << 63) : 0);
+}
+
+#endif
diff --git a/fs/erofs/fileio.c b/fs/erofs/fileio.c
new file mode 100644
index 000000000..ebb81a7ff
--- /dev/null
+++ b/fs/erofs/fileio.c
@@ -0,0 +1,217 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (C) 2024, Alibaba Cloud
+ */
+#include "internal.h"
+#include <trace/events/erofs.h>
+
+struct erofs_fileio_rq {
+	struct bio_vec bvecs[16];
+	struct bio bio;
+	struct kiocb iocb;
+	struct super_block *sb;
+	refcount_t ref;
+};
+
+struct erofs_fileio {
+	struct erofs_map_blocks map;
+	struct erofs_map_dev dev;
+	struct erofs_fileio_rq *rq;
+};
+
+static void erofs_fileio_ki_complete(struct kiocb *iocb, long ret)
+{
+	struct erofs_fileio_rq *rq =
+			container_of(iocb, struct erofs_fileio_rq, iocb);
+	struct folio_iter fi;
+
+	if (ret >= 0 && ret != rq->bio.bi_iter.bi_size)
+		ret = -EIO;
+	if (!rq->bio.bi_end_io) {
+		bio_for_each_folio_all(fi, &rq->bio) {
+			DBG_BUGON(folio_test_uptodate(fi.folio));
+			erofs_onlinefolio_end(fi.folio, ret < 0, false);
+		}
+	} else if (ret < 0 && !rq->bio.bi_status) {
+		rq->bio.bi_status = errno_to_blk_status(ret);
+	}
+	bio_endio(&rq->bio);
+	bio_uninit(&rq->bio);
+	if (refcount_dec_and_test(&rq->ref))
+		kfree(rq);
+}
+
+static void erofs_fileio_rq_submit(struct erofs_fileio_rq *rq)
+{
+	struct iov_iter iter;
+	ssize_t ret;
+
+	if (!rq)
+		return;
+	rq->iocb.ki_pos = rq->bio.bi_iter.bi_sector << SECTOR_SHIFT;
+	rq->iocb.ki_ioprio = get_current_ioprio();
+	rq->iocb.ki_complete = erofs_fileio_ki_complete;
+	if (test_opt(&EROFS_SB(rq->sb)->opt, DIRECT_IO) &&
+	    rq->iocb.ki_filp->f_mode & FMODE_CAN_ODIRECT)
+		rq->iocb.ki_flags = IOCB_DIRECT;
+	iov_iter_bvec(&iter, ITER_DEST, rq->bvecs, rq->bio.bi_vcnt,
+		      rq->bio.bi_iter.bi_size);
+	scoped_with_creds(rq->iocb.ki_filp->f_cred)
+		ret = vfs_iocb_iter_read(rq->iocb.ki_filp, &rq->iocb, &iter);
+	if (ret != -EIOCBQUEUED)
+		erofs_fileio_ki_complete(&rq->iocb, ret);
+	if (refcount_dec_and_test(&rq->ref))
+		kfree(rq);
+}
+
+static struct erofs_fileio_rq *erofs_fileio_rq_alloc(struct erofs_map_dev *mdev)
+{
+	struct erofs_fileio_rq *rq = kzalloc_obj(*rq, GFP_KERNEL | __GFP_NOFAIL);
+
+	bio_init(&rq->bio, NULL, rq->bvecs, ARRAY_SIZE(rq->bvecs), REQ_OP_READ);
+	rq->iocb.ki_filp = mdev->m_dif->file;
+	rq->sb = mdev->m_sb;
+	refcount_set(&rq->ref, 2);
+	return rq;
+}
+
+struct bio *erofs_fileio_bio_alloc(struct erofs_map_dev *mdev)
+{
+	return &erofs_fileio_rq_alloc(mdev)->bio;
+}
+
+void erofs_fileio_submit_bio(struct bio *bio)
+{
+	return erofs_fileio_rq_submit(container_of(bio, struct erofs_fileio_rq,
+						   bio));
+}
+
+static int erofs_fileio_scan_folio(struct erofs_fileio *io,
+				   struct inode *inode, struct folio *folio)
+{
+	struct erofs_map_blocks *map = &io->map;
+	unsigned int cur = 0, end = folio_size(folio), len, attached = 0;
+	loff_t pos = folio_pos(folio), ofs;
+	int err = 0;
+
+	erofs_onlinefolio_init(folio);
+	while (cur < end) {
+		if (!in_range(pos + cur, map->m_la, map->m_llen)) {
+			map->m_la = pos + cur;
+			map->m_llen = end - cur;
+			err = erofs_map_blocks(inode, map);
+			if (err)
+				break;
+		}
+
+		ofs = folio_pos(folio) + cur - map->m_la;
+		len = min_t(loff_t, map->m_llen - ofs, end - cur);
+		if (map->m_flags & EROFS_MAP_META) {
+			struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
+			void *src;
+
+			src = erofs_read_metabuf(&buf, inode->i_sb,
+				map->m_pa + ofs, erofs_inode_in_metabox(inode));
+			if (IS_ERR(src)) {
+				err = PTR_ERR(src);
+				break;
+			}
+			memcpy_to_folio(folio, cur, src, len);
+			erofs_put_metabuf(&buf);
+		} else if (!(map->m_flags & EROFS_MAP_MAPPED)) {
+			folio_zero_segment(folio, cur, cur + len);
+			attached = 0;
+		} else {
+			if (io->rq && (map->m_pa + ofs != io->dev.m_pa ||
+				       map->m_deviceid != io->dev.m_deviceid)) {
+io_retry:
+				erofs_fileio_rq_submit(io->rq);
+				io->rq = NULL;
+			}
+
+			if (!io->rq) {
+				io->dev = (struct erofs_map_dev) {
+					.m_pa = io->map.m_pa + ofs,
+					.m_deviceid = io->map.m_deviceid,
+				};
+				err = erofs_map_dev(inode->i_sb, &io->dev);
+				if (err)
+					break;
+				io->rq = erofs_fileio_rq_alloc(&io->dev);
+				io->rq->bio.bi_iter.bi_sector =
+					(io->dev.m_dif->fsoff + io->dev.m_pa) >> 9;
+				attached = 0;
+			}
+			if (!bio_add_folio(&io->rq->bio, folio, len, cur))
+				goto io_retry;
+			if (!attached++)
+				erofs_onlinefolio_split(folio);
+			io->dev.m_pa += len;
+		}
+		cur += len;
+	}
+	erofs_onlinefolio_end(folio, err, false);
+	return err;
+}
+
+static int erofs_fileio_read_folio(struct file *file, struct folio *folio)
+{
+	bool need_iput;
+	struct inode *realinode = erofs_real_inode(folio_inode(folio), &need_iput);
+	struct erofs_fileio io = {};
+	int err;
+
+	trace_erofs_read_folio(realinode, folio, true);
+	err = erofs_fileio_scan_folio(&io, realinode, folio);
+	erofs_fileio_rq_submit(io.rq);
+	if (need_iput)
+		iput(realinode);
+	return err;
+}
+
+static void erofs_fileio_readahead(struct readahead_control *rac)
+{
+	bool need_iput;
+	struct inode *realinode = erofs_real_inode(rac->mapping->host, &need_iput);
+	struct erofs_fileio io = {};
+	struct folio *folio;
+	int err;
+
+	trace_erofs_readahead(realinode, readahead_index(rac),
+			      readahead_count(rac), true);
+	while ((folio = readahead_folio(rac))) {
+		err = erofs_fileio_scan_folio(&io, realinode, folio);
+		if (err && err != -EINTR)
+			erofs_err(realinode->i_sb, "readahead error at folio %lu @ nid %llu",
+				  folio->index, EROFS_I(realinode)->nid);
+	}
+	erofs_fileio_rq_submit(io.rq);
+	if (need_iput)
+		iput(realinode);
+}
+
+const struct address_space_operations erofs_fileio_aops = {
+	.read_folio = erofs_fileio_read_folio,
+	.readahead = erofs_fileio_readahead,
+};
+
+int erofs_read_meta_folio(struct file *file, struct folio *folio)
+{
+	struct erofs_fileio io = {
+		.dev = { .m_pa = folio_pos(folio), },
+	};
+	struct inode *inode = folio_inode(folio);
+	int err;
+
+	err = erofs_map_dev(inode->i_sb, &io.dev);
+	if (err)
+		return err;
+
+	io.rq = erofs_fileio_rq_alloc(&io.dev);
+	io.rq->bio.bi_iter.bi_sector =
+		(io.dev.m_dif->fsoff + io.dev.m_pa) >> 9;
+	erofs_onlinefolio_init(folio);
+	bio_add_folio_nofail(&io.rq->bio, folio, folio_size(folio), 0);
+	erofs_fileio_rq_submit(io.rq);
+	return 0;
+}
diff --git a/fs/erofs/inode.c b/fs/erofs/inode.c
new file mode 100644
index 000000000..45afe5c50
--- /dev/null
+++ b/fs/erofs/inode.c
@@ -0,0 +1,399 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2017-2018 HUAWEI, Inc.
+ *             https://www.huawei.com/
+ * Copyright (C) 2021, Alibaba Cloud
+ */
+#include "xattr.h"
+#include <linux/compat.h>
+#include <trace/events/erofs.h>
+
+static int erofs_fill_symlink(struct inode *inode, void *bptr, unsigned int ofs)
+{
+	struct erofs_inode *vi = EROFS_I(inode);
+	char *link;
+	loff_t end;
+
+	ofs += vi->xattr_isize;
+	/* check whether the symlink data is small enough to be inlined */
+	if (vi->datalayout == EROFS_INODE_FLAT_INLINE &&
+	    !check_add_overflow(ofs, inode->i_size, &end) &&
+	    end <= i_blocksize(inode)) {
+		link = kmemdup_nul(bptr + ofs, inode->i_size, GFP_KERNEL);
+		if (!link)
+			return -ENOMEM;
+		if (unlikely(!inode->i_size || strlen(link) != inode->i_size)) {
+			erofs_err(inode->i_sb, "invalid fast symlink size %llu @ nid %llu",
+				  inode->i_size | 0ULL, vi->nid);
+			kfree(link);
+			return -EFSCORRUPTED;
+		}
+		inode_set_cached_link(inode, link, inode->i_size);
+	}
+	return 0;
+}
+
+static int erofs_read_inode(struct inode *inode)
+{
+	struct super_block *sb = inode->i_sb;
+	erofs_blk_t blkaddr = erofs_blknr(sb, erofs_iloc(inode));
+	unsigned int ofs = erofs_blkoff(sb, erofs_iloc(inode));
+	bool in_mbox = erofs_inode_in_metabox(inode);
+	struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
+	struct erofs_sb_info *sbi = EROFS_SB(sb);
+	erofs_blk_t addrmask = BIT_ULL(48) - 1;
+	struct erofs_inode *vi = EROFS_I(inode);
+	struct erofs_inode_extended *die, copied;
+	struct erofs_inode_compact *dic;
+	unsigned int ifmt;
+	void *ptr;
+	int err = 0;
+
+	ptr = erofs_read_metabuf(&buf, sb, erofs_pos(sb, blkaddr), in_mbox);
+	if (IS_ERR(ptr)) {
+		err = PTR_ERR(ptr);
+		erofs_err(sb, "failed to read inode meta block (nid: %llu): %d",
+			  vi->nid, err);
+		goto err_out;
+	}
+
+	dic = ptr + ofs;
+	ifmt = le16_to_cpu(dic->i_format);
+	if (ifmt & ~EROFS_I_ALL) {
+		erofs_err(sb, "unsupported i_format %u of nid %llu",
+			  ifmt, vi->nid);
+		err = -EOPNOTSUPP;
+		goto err_out;
+	}
+
+	vi->datalayout = erofs_inode_datalayout(ifmt);
+	if (vi->datalayout >= EROFS_INODE_DATALAYOUT_MAX) {
+		erofs_err(sb, "unsupported datalayout %u of nid %llu",
+			  vi->datalayout, vi->nid);
+		err = -EOPNOTSUPP;
+		goto err_out;
+	}
+
+	switch (erofs_inode_version(ifmt)) {
+	case EROFS_INODE_LAYOUT_EXTENDED:
+		vi->inode_isize = sizeof(struct erofs_inode_extended);
+		/* check if the extended inode acrosses block boundary */
+		if (ofs + vi->inode_isize <= sb->s_blocksize) {
+			ofs += vi->inode_isize;
+			die = (struct erofs_inode_extended *)dic;
+			copied.i_u = die->i_u;
+			copied.i_nb = die->i_nb;
+		} else {
+			const unsigned int gotten = sb->s_blocksize - ofs;
+
+			memcpy(&copied, dic, gotten);
+			ptr = erofs_read_metabuf(&buf, sb,
+					erofs_pos(sb, blkaddr + 1), in_mbox);
+			if (IS_ERR(ptr)) {
+				err = PTR_ERR(ptr);
+				erofs_err(sb, "failed to read inode payload block (nid: %llu): %d",
+					  vi->nid, err);
+				goto err_out;
+			}
+			ofs = vi->inode_isize - gotten;
+			memcpy((u8 *)&copied + gotten, ptr, ofs);
+			die = &copied;
+		}
+		vi->xattr_isize = erofs_xattr_ibody_size(die->i_xattr_icount);
+
+		inode->i_mode = le16_to_cpu(die->i_mode);
+		i_uid_write(inode, le32_to_cpu(die->i_uid));
+		i_gid_write(inode, le32_to_cpu(die->i_gid));
+		set_nlink(inode, le32_to_cpu(die->i_nlink));
+		inode_set_mtime(inode, le64_to_cpu(die->i_mtime),
+				le32_to_cpu(die->i_mtime_nsec));
+
+		inode->i_size = le64_to_cpu(die->i_size);
+		break;
+	case EROFS_INODE_LAYOUT_COMPACT:
+		vi->inode_isize = sizeof(struct erofs_inode_compact);
+		ofs += vi->inode_isize;
+		vi->xattr_isize = erofs_xattr_ibody_size(dic->i_xattr_icount);
+
+		inode->i_mode = le16_to_cpu(dic->i_mode);
+		copied.i_u = dic->i_u;
+		i_uid_write(inode, le16_to_cpu(dic->i_uid));
+		i_gid_write(inode, le16_to_cpu(dic->i_gid));
+		if (!S_ISDIR(inode->i_mode) &&
+		    ((ifmt >> EROFS_I_NLINK_1_BIT) & 1)) {
+			set_nlink(inode, 1);
+			copied.i_nb = dic->i_nb;
+		} else {
+			set_nlink(inode, le16_to_cpu(dic->i_nb.nlink));
+			copied.i_nb.startblk_hi = 0;
+			addrmask = BIT_ULL(32) - 1;
+		}
+		inode_set_mtime(inode, sbi->epoch + le32_to_cpu(dic->i_mtime),
+				sbi->fixed_nsec);
+
+		inode->i_size = le32_to_cpu(dic->i_size);
+		break;
+	default:
+		erofs_err(sb, "unsupported on-disk inode version %u of nid %llu",
+			  erofs_inode_version(ifmt), vi->nid);
+		err = -EOPNOTSUPP;
+		goto err_out;
+	}
+
+	if (unlikely(inode->i_size < 0)) {
+		erofs_err(sb, "negative i_size @ nid %llu", vi->nid);
+		err = -EFSCORRUPTED;
+		goto err_out;
+	}
+
+	if (IS_ENABLED(CONFIG_EROFS_FS_POSIX_ACL) &&
+	    erofs_inode_has_noacl(inode, ptr, ofs))
+		cache_no_acl(inode);
+
+	switch (inode->i_mode & S_IFMT) {
+	case S_IFDIR:
+		vi->dot_omitted = (ifmt >> EROFS_I_DOT_OMITTED_BIT) & 1;
+		fallthrough;
+	case S_IFREG:
+	case S_IFLNK:
+		vi->startblk = le32_to_cpu(copied.i_u.startblk_lo) |
+			((u64)le16_to_cpu(copied.i_nb.startblk_hi) << 32);
+		if (vi->datalayout == EROFS_INODE_FLAT_PLAIN &&
+		    !((vi->startblk ^ EROFS_NULL_ADDR) & addrmask))
+			vi->startblk = EROFS_NULL_ADDR;
+
+		if(S_ISLNK(inode->i_mode)) {
+			err = erofs_fill_symlink(inode, ptr, ofs);
+			if (err)
+				goto err_out;
+		}
+		break;
+	case S_IFCHR:
+	case S_IFBLK:
+		inode->i_rdev = new_decode_dev(le32_to_cpu(copied.i_u.rdev));
+		break;
+	case S_IFIFO:
+	case S_IFSOCK:
+		inode->i_rdev = 0;
+		break;
+	default:
+		erofs_err(sb, "bogus i_mode (%o) @ nid %llu", inode->i_mode,
+			  vi->nid);
+		err = -EFSCORRUPTED;
+		goto err_out;
+	}
+
+	if (!erofs_inode_is_data_compressed(vi->datalayout)) {
+		inode->i_blocks = round_up(inode->i_size, sb->s_blocksize) >> 9;
+	} else if (!IS_ENABLED(CONFIG_EROFS_FS_ZIP) || !sbi->available_compr_algs) {
+		erofs_err(sb, "compressed inode (nid %llu) is invalid in a plain filesystem",
+			  vi->nid);
+		err = -EFSCORRUPTED;
+		goto err_out;
+	} else {
+		inode->i_blocks = (le32_to_cpu(copied.i_u.blocks_lo) |
+			((u64)le16_to_cpu(copied.i_nb.blocks_hi) << 32)) <<
+				  (sb->s_blocksize_bits - 9);
+	}
+
+	if (vi->datalayout == EROFS_INODE_CHUNK_BASED) {
+		/* fill chunked inode summary info */
+		vi->chunkformat = le16_to_cpu(copied.i_u.c.format);
+		if (vi->chunkformat & ~EROFS_CHUNK_FORMAT_ALL) {
+			erofs_err(sb, "unsupported chunk format %x of nid %llu",
+				  vi->chunkformat, vi->nid);
+			err = -EOPNOTSUPP;
+			goto err_out;
+		}
+		vi->chunkbits = sb->s_blocksize_bits +
+			(vi->chunkformat & EROFS_CHUNK_FORMAT_BLKBITS_MASK);
+	}
+	inode_set_atime_to_ts(inode,
+			      inode_set_ctime_to_ts(inode, inode_get_mtime(inode)));
+
+	inode->i_flags &= ~S_DAX;
+	if (test_opt(&sbi->opt, DAX_ALWAYS) && S_ISREG(inode->i_mode) &&
+	    (vi->datalayout == EROFS_INODE_FLAT_PLAIN ||
+	     vi->datalayout == EROFS_INODE_CHUNK_BASED))
+		inode->i_flags |= S_DAX;
+err_out:
+	erofs_put_metabuf(&buf);
+	return err;
+}
+
+static int erofs_fill_inode(struct inode *inode)
+{
+	const struct address_space_operations *aops;
+	int err;
+
+	trace_erofs_fill_inode(inode);
+	err = erofs_read_inode(inode);
+	if (err)
+		return err;
+
+	switch (inode->i_mode & S_IFMT) {
+	case S_IFREG:
+		inode->i_op = &erofs_generic_iops;
+		inode->i_fop = erofs_ishare_fill_inode(inode) ?
+			       &erofs_ishare_fops : &erofs_file_fops;
+		break;
+	case S_IFDIR:
+		inode->i_op = &erofs_dir_iops;
+		inode->i_fop = &erofs_dir_fops;
+		inode_nohighmem(inode);
+		break;
+	case S_IFLNK:
+		if (inode->i_link)
+			inode->i_op = &erofs_fast_symlink_iops;
+		else
+			inode->i_op = &erofs_symlink_iops;
+		inode_nohighmem(inode);
+		break;
+	default:
+		inode->i_op = &erofs_generic_iops;
+		init_special_inode(inode, inode->i_mode, inode->i_rdev);
+		return 0;
+	}
+
+	mapping_set_large_folios(inode->i_mapping);
+	aops = erofs_get_aops(inode);
+	if (IS_ERR(aops))
+		return PTR_ERR(aops);
+	inode->i_mapping->a_ops = aops;
+	return 0;
+}
+
+/*
+ * ino_t is 32-bits on 32-bit arch. We have to squash the 64-bit value down
+ * so that it will fit.
+ */
+static ino_t erofs_squash_ino(struct super_block *sb, erofs_nid_t nid)
+{
+	u64 ino64 = erofs_nid_to_ino64(EROFS_SB(sb), nid);
+
+	if (sizeof(ino_t) < sizeof(erofs_nid_t))
+		ino64 ^= ino64 >> (sizeof(erofs_nid_t) - sizeof(ino_t)) * 8;
+	return (ino_t)ino64;
+}
+
+static int erofs_iget5_eq(struct inode *inode, void *opaque)
+{
+	return EROFS_I(inode)->nid == *(erofs_nid_t *)opaque;
+}
+
+static int erofs_iget5_set(struct inode *inode, void *opaque)
+{
+	const erofs_nid_t nid = *(erofs_nid_t *)opaque;
+
+	inode->i_ino = erofs_squash_ino(inode->i_sb, nid);
+	EROFS_I(inode)->nid = nid;
+	return 0;
+}
+
+struct inode *erofs_iget(struct super_block *sb, erofs_nid_t nid)
+{
+	struct inode *inode;
+
+	inode = iget5_locked(sb, erofs_squash_ino(sb, nid), erofs_iget5_eq,
+			     erofs_iget5_set, &nid);
+	if (!inode)
+		return ERR_PTR(-ENOMEM);
+
+	if (inode_state_read_once(inode) & I_NEW) {
+		int err = erofs_fill_inode(inode);
+
+		if (err) {
+			iget_failed(inode);
+			return ERR_PTR(err);
+		}
+		unlock_new_inode(inode);
+	}
+	return inode;
+}
+
+int erofs_getattr(struct mnt_idmap *idmap, const struct path *path,
+		  struct kstat *stat, u32 request_mask,
+		  unsigned int query_flags)
+{
+	struct inode *const inode = d_inode(path->dentry);
+	struct block_device *bdev = inode->i_sb->s_bdev;
+	bool compressed =
+		erofs_inode_is_data_compressed(EROFS_I(inode)->datalayout);
+
+	if (compressed)
+		stat->attributes |= STATX_ATTR_COMPRESSED;
+	stat->attributes |= STATX_ATTR_IMMUTABLE;
+	stat->attributes_mask |= (STATX_ATTR_COMPRESSED |
+				  STATX_ATTR_IMMUTABLE);
+
+	/*
+	 * Return the DIO alignment restrictions if requested.
+	 *
+	 * In EROFS, STATX_DIOALIGN is only supported in bdev-based mode
+	 * and uncompressed inodes, otherwise we report no DIO support.
+	 */
+	if ((request_mask & STATX_DIOALIGN) && S_ISREG(inode->i_mode)) {
+		stat->result_mask |= STATX_DIOALIGN;
+		if (bdev && !compressed) {
+			stat->dio_mem_align = bdev_dma_alignment(bdev) + 1;
+			stat->dio_offset_align = bdev_logical_block_size(bdev);
+		}
+	}
+	generic_fillattr(idmap, request_mask, inode, stat);
+	return 0;
+}
+
+static int erofs_ioctl_get_volume_label(struct inode *inode, void __user *arg)
+{
+	struct erofs_sb_info *sbi = EROFS_I_SB(inode);
+	int ret;
+
+	if (!sbi->volume_name)
+		ret = clear_user(arg, 1);
+	else
+		ret = copy_to_user(arg, sbi->volume_name,
+				   strlen(sbi->volume_name) + 1);
+	return ret ? -EFAULT : 0;
+}
+
+long erofs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
+{
+	struct inode *inode = file_inode(filp);
+	void __user *argp = (void __user *)arg;
+
+	switch (cmd) {
+	case FS_IOC_GETFSLABEL:
+		return erofs_ioctl_get_volume_label(inode, argp);
+	default:
+		return -ENOTTY;
+	}
+}
+
+#ifdef CONFIG_COMPAT
+long erofs_compat_ioctl(struct file *filp, unsigned int cmd,
+			unsigned long arg)
+{
+	return erofs_ioctl(filp, cmd, (unsigned long)compat_ptr(arg));
+}
+#endif
+
+const struct inode_operations erofs_generic_iops = {
+	.getattr = erofs_getattr,
+	.listxattr = erofs_listxattr,
+	.get_inode_acl = erofs_get_acl,
+	.fiemap = erofs_fiemap,
+};
+
+const struct inode_operations erofs_symlink_iops = {
+	.get_link = page_get_link,
+	.getattr = erofs_getattr,
+	.listxattr = erofs_listxattr,
+	.get_inode_acl = erofs_get_acl,
+};
+
+const struct inode_operations erofs_fast_symlink_iops = {
+	.get_link = simple_get_link,
+	.getattr = erofs_getattr,
+	.listxattr = erofs_listxattr,
+	.get_inode_acl = erofs_get_acl,
+};
diff --git a/fs/erofs/internal.h b/fs/erofs/internal.h
new file mode 100644
index 000000000..9b7370f0f
--- /dev/null
+++ b/fs/erofs/internal.h
@@ -0,0 +1,539 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (C) 2017-2018 HUAWEI, Inc.
+ *             https://www.huawei.com/
+ * Copyright (C) 2021, Alibaba Cloud
+ */
+#ifndef __EROFS_INTERNAL_H
+#define __EROFS_INTERNAL_H
+
+#include <linux/fs.h>
+#include <linux/dax.h>
+#include <linux/dcache.h>
+#include <linux/mm.h>
+#include <linux/module.h>
+#include <linux/pagemap.h>
+#include <linux/bio.h>
+#include <linux/magic.h>
+#include <linux/slab.h>
+#include <linux/vmalloc.h>
+#include <linux/iomap.h>
+#include "erofs_fs.h"
+
+__printf(2, 3) void _erofs_printk(struct super_block *sb, const char *fmt, ...);
+#define erofs_err(sb, fmt, ...)	\
+	_erofs_printk(sb, KERN_ERR fmt "\n", ##__VA_ARGS__)
+#define erofs_warn(sb, fmt, ...) \
+	_erofs_printk(sb, KERN_WARNING fmt "\n", ##__VA_ARGS__)
+#define erofs_info(sb, fmt, ...) \
+	_erofs_printk(sb, KERN_INFO fmt "\n", ##__VA_ARGS__)
+
+#ifdef CONFIG_EROFS_FS_DEBUG
+#define DBG_BUGON               BUG_ON
+#else
+#define DBG_BUGON(x)            ((void)(x))
+#endif	/* !CONFIG_EROFS_FS_DEBUG */
+
+/* EROFS_SUPER_MAGIC_V1 to represent the whole file system */
+#define EROFS_SUPER_MAGIC   EROFS_SUPER_MAGIC_V1
+
+typedef u64 erofs_nid_t;
+typedef u64 erofs_off_t;
+typedef u64 erofs_blk_t;
+
+struct erofs_device_info {
+	char *path;
+	struct file *file;
+	struct dax_device *dax_dev;
+	u64 fsoff, dax_part_off;
+
+	erofs_blk_t blocks;
+	erofs_blk_t uniaddr;
+};
+
+enum {
+	EROFS_SYNC_DECOMPRESS_AUTO,
+	EROFS_SYNC_DECOMPRESS_FORCE_ON,
+	EROFS_SYNC_DECOMPRESS_FORCE_OFF
+};
+
+struct erofs_mount_opts {
+	/* current strategy of how to use managed cache */
+	unsigned char cache_strategy;
+	unsigned int mount_opt;
+};
+
+struct erofs_dev_context {
+	struct idr tree;
+	struct rw_semaphore rwsem;
+
+	unsigned int extra_devices;
+	bool flatdev;
+};
+
+/* all filesystem-wide lz4 configurations */
+struct erofs_sb_lz4_info {
+	/* # of pages needed for EROFS lz4 rolling decompression */
+	u16 max_distance_pages;
+	/* maximum possible blocks for pclusters in the filesystem */
+	u16 max_pclusterblks;
+};
+
+struct erofs_xattr_prefix_item {
+	struct erofs_xattr_long_prefix *prefix;
+	u8 infix_len;
+};
+
+struct erofs_sb_info {
+	struct erofs_device_info dif0;
+	struct erofs_mount_opts opt;	/* options */
+#ifdef CONFIG_EROFS_FS_ZIP
+	/* list for all registered superblocks, mainly for shrinker */
+	struct list_head list;
+	struct mutex umount_mutex;
+
+	/* managed XArray arranged in physical block number */
+	struct xarray managed_pslots;
+
+	unsigned int sync_decompress;	/* strategy for sync decompression */
+	unsigned int shrinker_run_no;
+
+	struct erofs_sb_lz4_info lz4;
+#endif	/* CONFIG_EROFS_FS_ZIP */
+	struct inode *managed_cache; /* pseudo inode to cache physical data */
+	struct inode *packed_inode;
+	struct inode *metabox_inode;
+	struct erofs_dev_context *devs;
+	u64 total_blocks;
+
+	u32 meta_blkaddr;
+#ifdef CONFIG_EROFS_FS_XATTR
+	u32 xattr_blkaddr;
+	u32 xattr_prefix_start;
+	u8 xattr_prefix_count;
+	u8 ishare_xattr_prefix_id;
+	struct erofs_xattr_prefix_item *xattr_prefixes;
+	unsigned int xattr_filter_reserved;
+#endif
+	u16 device_id_mask;	/* valid bits of device id to be used */
+
+	unsigned char islotbits;	/* inode slot unit size in bit shift */
+	unsigned char blkszbits;	/* filesystem block size in bit shift */
+
+	u32 sb_size;			/* total superblock size */
+	u32 fixed_nsec;
+	s64 epoch;
+
+	/* what we really care is nid, rather than ino.. */
+	erofs_nid_t root_nid;
+	erofs_nid_t packed_nid;
+	erofs_nid_t metabox_nid;
+	/* used for statfs, f_files - f_favail */
+	u64 inos;
+
+	char *volume_name;
+	u32 feature_compat;
+	u32 feature_incompat;
+	u16 available_compr_algs;
+
+	/* sysfs support */
+	struct kobject s_kobj;		/* /sys/fs/erofs/<devname> */
+	struct completion s_kobj_unregister;
+	erofs_off_t dir_ra_bytes;
+
+	char *domain_id;
+};
+
+#define EROFS_SB(sb) ((struct erofs_sb_info *)(sb)->s_fs_info)
+#define EROFS_I_SB(inode) ((struct erofs_sb_info *)(inode)->i_sb->s_fs_info)
+
+/* Mount flags set via mount options or defaults */
+#define EROFS_MOUNT_XATTR_USER		0x00000010
+#define EROFS_MOUNT_POSIX_ACL		0x00000020
+#define EROFS_MOUNT_DAX_ALWAYS		0x00000040
+#define EROFS_MOUNT_DAX_NEVER		0x00000080
+#define EROFS_MOUNT_DIRECT_IO		0x00000100
+#define EROFS_MOUNT_INODE_SHARE		0x00000200
+
+#define clear_opt(opt, option)	((opt)->mount_opt &= ~EROFS_MOUNT_##option)
+#define set_opt(opt, option)	((opt)->mount_opt |= EROFS_MOUNT_##option)
+#define test_opt(opt, option)	((opt)->mount_opt & EROFS_MOUNT_##option)
+
+static inline bool erofs_is_fileio_mode(struct erofs_sb_info *sbi)
+{
+	return IS_ENABLED(CONFIG_EROFS_FS_BACKED_BY_FILE) && sbi->dif0.file;
+}
+
+extern struct file_system_type erofs_anon_fs_type;
+
+enum {
+	EROFS_ZIP_CACHE_DISABLED,
+	EROFS_ZIP_CACHE_READAHEAD,
+	EROFS_ZIP_CACHE_READAROUND
+};
+
+struct erofs_buf {
+	struct address_space *mapping;
+	u64 off;
+	struct page *page;
+	void *base;
+	bool mc;
+};
+#define __EROFS_BUF_INITIALIZER	((struct erofs_buf){ .page = NULL })
+
+#define erofs_blknr(sb, pos)	((erofs_blk_t)((pos) >> (sb)->s_blocksize_bits))
+#define erofs_blkoff(sb, pos)	((pos) & ((sb)->s_blocksize - 1))
+#define erofs_pos(sb, blk)	((erofs_off_t)(blk) << (sb)->s_blocksize_bits)
+#define erofs_iblks(i)	(round_up((i)->i_size, i_blocksize(i)) >> (i)->i_blkbits)
+
+#define EROFS_FEATURE_FUNCS(name, compat, feature) \
+static inline bool erofs_sb_has_##name(struct erofs_sb_info *sbi) \
+{ \
+	return sbi->feature_##compat & EROFS_FEATURE_##feature; \
+}
+
+EROFS_FEATURE_FUNCS(lz4_0padding, incompat, INCOMPAT_LZ4_0PADDING)
+EROFS_FEATURE_FUNCS(compr_cfgs, incompat, INCOMPAT_COMPR_CFGS)
+EROFS_FEATURE_FUNCS(big_pcluster, incompat, INCOMPAT_BIG_PCLUSTER)
+EROFS_FEATURE_FUNCS(chunked_file, incompat, INCOMPAT_CHUNKED_FILE)
+EROFS_FEATURE_FUNCS(device_table, incompat, INCOMPAT_DEVICE_TABLE)
+EROFS_FEATURE_FUNCS(compr_head2, incompat, INCOMPAT_COMPR_HEAD2)
+EROFS_FEATURE_FUNCS(ztailpacking, incompat, INCOMPAT_ZTAILPACKING)
+EROFS_FEATURE_FUNCS(fragments, incompat, INCOMPAT_FRAGMENTS)
+EROFS_FEATURE_FUNCS(dedupe, incompat, INCOMPAT_DEDUPE)
+EROFS_FEATURE_FUNCS(xattr_prefixes, incompat, INCOMPAT_XATTR_PREFIXES)
+EROFS_FEATURE_FUNCS(48bit, incompat, INCOMPAT_48BIT)
+EROFS_FEATURE_FUNCS(metabox, incompat, INCOMPAT_METABOX)
+EROFS_FEATURE_FUNCS(sb_chksum, compat, COMPAT_SB_CHKSUM)
+EROFS_FEATURE_FUNCS(xattr_filter, compat, COMPAT_XATTR_FILTER)
+EROFS_FEATURE_FUNCS(shared_ea_in_metabox, compat, COMPAT_SHARED_EA_IN_METABOX)
+EROFS_FEATURE_FUNCS(plain_xattr_pfx, compat, COMPAT_PLAIN_XATTR_PFX)
+EROFS_FEATURE_FUNCS(ishare_xattrs, compat, COMPAT_ISHARE_XATTRS)
+
+static inline u64 erofs_nid_to_ino64(struct erofs_sb_info *sbi, erofs_nid_t nid)
+{
+	if (!erofs_sb_has_metabox(sbi))
+		return nid;
+
+	/*
+	 * When metadata compression is enabled, avoid generating excessively
+	 * large inode numbers for metadata-compressed inodes.  Shift NIDs in
+	 * the 31-62 bit range left by one and move the metabox flag to bit 31.
+	 *
+	 * Note: on-disk NIDs remain unchanged as they are primarily used for
+	 * compatibility with non-LFS 32-bit applications.
+	 */
+	return ((nid << 1) & GENMASK_ULL(63, 32)) | (nid & GENMASK(30, 0)) |
+		((nid >> EROFS_DIRENT_NID_METABOX_BIT) << 31);
+}
+
+/* atomic flag definitions */
+#define EROFS_I_EA_INITED_BIT	0
+#define EROFS_I_Z_INITED_BIT	1
+
+/* bitlock definitions (arranged in reverse order) */
+#define EROFS_I_BL_XATTR_BIT	(BITS_PER_LONG - 1)
+#define EROFS_I_BL_Z_BIT	(BITS_PER_LONG - 2)
+
+/* default readahead size of directories */
+#define EROFS_DIR_RA_BYTES	16384
+
+struct erofs_inode_fingerprint {
+	u8 *opaque;
+	int size;
+};
+
+struct erofs_inode {
+	erofs_nid_t nid;
+
+	/* atomic flags (including bitlocks) */
+	unsigned long flags;
+
+	unsigned char datalayout;
+	unsigned char inode_isize;
+	bool dot_omitted;
+	unsigned int xattr_isize;
+
+	unsigned int xattr_name_filter;
+	unsigned int xattr_shared_count;
+	unsigned int *xattr_shared_xattrs;
+
+	union {
+		erofs_blk_t startblk;
+		struct {
+			unsigned short	chunkformat;
+			unsigned char	chunkbits;
+		};
+#ifdef CONFIG_EROFS_FS_ZIP
+		struct {
+			unsigned short z_advise;
+			unsigned char  z_algofmt[2];
+			unsigned char  z_lclusterbits;
+			union {
+				u64    z_tailextent_headlcn;
+				u64    z_extents;
+			};
+			erofs_off_t    z_fragmentoff;
+			unsigned short z_idata_size;
+		};
+#endif	/* CONFIG_EROFS_FS_ZIP */
+	};
+#ifdef CONFIG_EROFS_FS_PAGE_CACHE_SHARE
+	struct list_head ishare_list;
+	union {
+		/* for each anon shared inode */
+		struct {
+			struct erofs_inode_fingerprint fingerprint;
+			spinlock_t ishare_lock;
+		};
+		/* for each real filesystem inode */
+		struct dentry *sharedentry;
+	};
+#endif
+	/* the corresponding vfs inode */
+	struct inode vfs_inode;
+};
+
+#define EROFS_I(ptr)	container_of(ptr, struct erofs_inode, vfs_inode)
+
+static inline bool erofs_inode_in_metabox(struct inode *inode)
+{
+	return EROFS_I(inode)->nid & BIT_ULL(EROFS_DIRENT_NID_METABOX_BIT);
+}
+
+static inline erofs_off_t erofs_iloc(struct inode *inode)
+{
+	struct erofs_sb_info *sbi = EROFS_I_SB(inode);
+	erofs_nid_t nid_lo = EROFS_I(inode)->nid & EROFS_DIRENT_NID_MASK;
+
+	if (erofs_inode_in_metabox(inode))
+		return nid_lo << sbi->islotbits;
+	return erofs_pos(inode->i_sb, sbi->meta_blkaddr) +
+		(nid_lo << sbi->islotbits);
+}
+
+static inline unsigned int erofs_inode_version(unsigned int ifmt)
+{
+	return (ifmt >> EROFS_I_VERSION_BIT) & EROFS_I_VERSION_MASK;
+}
+
+static inline unsigned int erofs_inode_datalayout(unsigned int ifmt)
+{
+	return (ifmt >> EROFS_I_DATALAYOUT_BIT) & EROFS_I_DATALAYOUT_MASK;
+}
+
+/* reclaiming is never triggered when allocating new folios. */
+static inline struct folio *erofs_grab_folio_nowait(struct address_space *as,
+						    pgoff_t index)
+{
+	return __filemap_get_folio(as, index,
+			FGP_LOCK|FGP_CREAT|FGP_NOFS|FGP_NOWAIT,
+			readahead_gfp_mask(as) & ~__GFP_RECLAIM);
+}
+
+/* Allocated on disk at @m_pa (e.g. NOT a fragment extent) */
+#define EROFS_MAP_MAPPED		0x0001
+/* Located in metadata (could be copied from bd_inode) */
+#define EROFS_MAP_META			0x0002
+/* @m_llen may be truncated by the runtime compared to the on-disk record */
+#define EROFS_MAP_PARTIAL_MAPPED	0x0004
+/* The on-disk @m_llen may cover only part of the encoded data */
+#define EROFS_MAP_PARTIAL_REF		0x0008
+/* Located in the special packed inode */
+#define EROFS_MAP_FRAGMENT		0x0010
+/* The encoded on-disk data will be fully handled (decompressed) */
+#define EROFS_MAP_FULL(f)	(!((f) & (EROFS_MAP_PARTIAL_MAPPED | \
+					  EROFS_MAP_PARTIAL_REF)))
+
+struct erofs_map_blocks {
+	struct erofs_buf buf;
+
+	erofs_off_t m_pa, m_la;
+	u64 m_plen, m_llen;
+
+	unsigned short m_deviceid;
+	char m_algorithmformat;
+	unsigned int m_flags;
+};
+
+/*
+ * Used to get the exact decompressed length, e.g. fiemap (consider lookback
+ * approach instead if possible since it's more metadata lightweight.)
+ */
+#define EROFS_GET_BLOCKS_FIEMAP		0x0001
+/* Used to map the whole extent if non-negligible data is requested for LZMA */
+#define EROFS_GET_BLOCKS_READMORE	0x0002
+/* Used to map tail extent for tailpacking inline or fragment pcluster */
+#define EROFS_GET_BLOCKS_FINDTAIL	0x0004
+
+enum {
+	Z_EROFS_COMPRESSION_SHIFTED = Z_EROFS_COMPRESSION_MAX,
+	Z_EROFS_COMPRESSION_INTERLACED,
+	Z_EROFS_COMPRESSION_RUNTIME_MAX
+};
+
+struct erofs_map_dev {
+	struct super_block *m_sb;
+	struct erofs_device_info *m_dif;
+	struct block_device *m_bdev;
+
+	erofs_off_t m_pa;
+	unsigned int m_deviceid;
+};
+
+extern const struct super_operations erofs_sops;
+extern const struct address_space_operations erofs_aops;
+extern const struct address_space_operations erofs_fileio_aops;
+extern const struct address_space_operations z_erofs_aops;
+
+extern const struct inode_operations erofs_generic_iops;
+extern const struct inode_operations erofs_symlink_iops;
+extern const struct inode_operations erofs_fast_symlink_iops;
+extern const struct inode_operations erofs_dir_iops;
+
+extern const struct file_operations erofs_file_fops;
+extern const struct file_operations erofs_dir_fops;
+extern const struct file_operations erofs_ishare_fops;
+
+extern const struct iomap_ops z_erofs_iomap_report_ops;
+
+int erofs_setup_managed_cache(struct super_block *sb);
+#ifdef CONFIG_EROFS_FS_BACKED_BY_FILE
+int erofs_read_meta_folio(struct file *file, struct folio *folio);
+#else
+#define erofs_read_meta_folio NULL
+#endif
+void *erofs_read_metadata(struct super_block *sb, struct erofs_buf *buf,
+			  erofs_off_t *offset, int *lengthp);
+void erofs_unmap_metabuf(struct erofs_buf *buf);
+void erofs_put_metabuf(struct erofs_buf *buf);
+void *erofs_bread(struct erofs_buf *buf, erofs_off_t offset, bool need_kmap);
+int erofs_init_metabuf(struct erofs_buf *buf, struct super_block *sb,
+		       bool in_metabox);
+void *erofs_read_metabuf(struct erofs_buf *buf, struct super_block *sb,
+			 erofs_off_t offset, bool in_metabox);
+int erofs_map_dev(struct super_block *sb, struct erofs_map_dev *dev);
+int erofs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
+		 u64 start, u64 len);
+int erofs_map_blocks(struct inode *inode, struct erofs_map_blocks *map);
+void erofs_onlinefolio_init(struct folio *folio);
+void erofs_onlinefolio_split(struct folio *folio);
+void erofs_onlinefolio_end(struct folio *folio, int err, bool dirty);
+struct inode *erofs_iget(struct super_block *sb, erofs_nid_t nid);
+int erofs_getattr(struct mnt_idmap *idmap, const struct path *path,
+		  struct kstat *stat, u32 request_mask,
+		  unsigned int query_flags);
+int erofs_namei(struct inode *dir, const struct qstr *name,
+		erofs_nid_t *nid, unsigned int *d_type);
+
+static inline void *erofs_vm_map_ram(struct page **pages, unsigned int count)
+{
+	int retried = 0;
+
+	while (1) {
+		void *p = vm_map_ram(pages, count, -1);
+
+		/* retry two more times (totally 3 times) */
+		if (p || ++retried >= 3)
+			return p;
+		vm_unmap_aliases();
+	}
+	return NULL;
+}
+
+static inline const struct address_space_operations *
+erofs_get_aops(struct inode *realinode)
+{
+	if (erofs_inode_is_data_compressed(EROFS_I(realinode)->datalayout)) {
+		if (!IS_ENABLED(CONFIG_EROFS_FS_ZIP))
+			return ERR_PTR(-EOPNOTSUPP);
+		DO_ONCE_LITE_IF(realinode->i_blkbits != PAGE_SHIFT,
+			  erofs_info, realinode->i_sb,
+			  "EXPERIMENTAL EROFS subpage compressed block support in use. Use at your own risk!");
+		return &z_erofs_aops;
+	}
+	if (IS_ENABLED(CONFIG_EROFS_FS_BACKED_BY_FILE) &&
+	    erofs_is_fileio_mode(EROFS_SB(realinode->i_sb)))
+		return &erofs_fileio_aops;
+	return &erofs_aops;
+}
+
+int erofs_register_sysfs(struct super_block *sb);
+void erofs_unregister_sysfs(struct super_block *sb);
+int __init erofs_init_sysfs(void);
+void erofs_exit_sysfs(void);
+
+struct page *__erofs_allocpage(struct page **pagepool, gfp_t gfp, bool tryrsv);
+static inline struct page *erofs_allocpage(struct page **pagepool, gfp_t gfp)
+{
+	return __erofs_allocpage(pagepool, gfp, false);
+}
+static inline void erofs_pagepool_add(struct page **pagepool, struct page *page)
+{
+	set_page_private(page, (unsigned long)*pagepool);
+	*pagepool = page;
+}
+void erofs_release_pages(struct page **pagepool);
+
+#ifdef CONFIG_EROFS_FS_ZIP
+#define MNGD_MAPPING(sbi)	((sbi)->managed_cache->i_mapping)
+
+extern atomic_long_t erofs_global_shrink_cnt;
+void erofs_shrinker_register(struct super_block *sb);
+void erofs_shrinker_unregister(struct super_block *sb);
+int __init erofs_init_shrinker(void);
+void erofs_exit_shrinker(void);
+int __init z_erofs_init_subsystem(void);
+void z_erofs_exit_subsystem(void);
+int z_erofs_init_super(struct super_block *sb);
+unsigned long z_erofs_shrink_scan(struct erofs_sb_info *sbi,
+				  unsigned long nr_shrink);
+int z_erofs_map_blocks_iter(struct inode *inode, struct erofs_map_blocks *map,
+			    int flags);
+void *z_erofs_get_gbuf(unsigned int requiredpages);
+void z_erofs_put_gbuf(void *ptr);
+int z_erofs_gbuf_growsize(unsigned int nrpages);
+int __init z_erofs_gbuf_init(void);
+void z_erofs_gbuf_exit(void);
+#else
+static inline void erofs_shrinker_register(struct super_block *sb) {}
+static inline void erofs_shrinker_unregister(struct super_block *sb) {}
+static inline int erofs_init_shrinker(void) { return 0; }
+static inline void erofs_exit_shrinker(void) {}
+static inline int z_erofs_init_subsystem(void) { return 0; }
+static inline void z_erofs_exit_subsystem(void) {}
+static inline int z_erofs_init_super(struct super_block *sb) { return 0; }
+#endif	/* !CONFIG_EROFS_FS_ZIP */
+int z_erofs_parse_cfgs(struct super_block *sb, struct erofs_super_block *dsb);
+
+#ifdef CONFIG_EROFS_FS_BACKED_BY_FILE
+struct bio *erofs_fileio_bio_alloc(struct erofs_map_dev *mdev);
+void erofs_fileio_submit_bio(struct bio *bio);
+#else
+static inline struct bio *erofs_fileio_bio_alloc(struct erofs_map_dev *mdev) { return NULL; }
+static inline void erofs_fileio_submit_bio(struct bio *bio) {}
+#endif
+
+#ifdef CONFIG_EROFS_FS_PAGE_CACHE_SHARE
+int __init erofs_init_ishare(void);
+void erofs_exit_ishare(void);
+bool erofs_ishare_fill_inode(struct inode *inode);
+void erofs_ishare_free_inode(struct inode *inode);
+struct inode *erofs_real_inode(struct inode *inode, bool *need_iput);
+#else
+static inline int erofs_init_ishare(void) { return 0; }
+static inline void erofs_exit_ishare(void) {}
+static inline bool erofs_ishare_fill_inode(struct inode *inode) { return false; }
+static inline void erofs_ishare_free_inode(struct inode *inode) {}
+static inline struct inode *erofs_real_inode(struct inode *inode, bool *need_iput)
+{
+	*need_iput = false;
+	return inode;
+}
+#endif
+
+long erofs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
+long erofs_compat_ioctl(struct file *filp, unsigned int cmd,
+			unsigned long arg);
+
+#endif	/* __EROFS_INTERNAL_H */
diff --git a/fs/erofs/ishare.c b/fs/erofs/ishare.c
new file mode 100644
index 000000000..fa7d4112d
--- /dev/null
+++ b/fs/erofs/ishare.c
@@ -0,0 +1,214 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (C) 2024, Alibaba Cloud
+ */
+#include <linux/backing-file.h>
+#include <linux/xxhash.h>
+#include <linux/mount.h>
+#include <linux/security.h>
+#include "internal.h"
+#include "xattr.h"
+
+static struct vfsmount *erofs_ishare_mnt;
+
+static int erofs_ishare_iget5_eq(struct inode *inode, void *data)
+{
+	struct erofs_inode_fingerprint *fp1 = &EROFS_I(inode)->fingerprint;
+	struct erofs_inode_fingerprint *fp2 = data;
+
+	return fp1->size == fp2->size &&
+		!memcmp(fp1->opaque, fp2->opaque, fp2->size);
+}
+
+static int erofs_ishare_iget5_set(struct inode *inode, void *data)
+{
+	struct erofs_inode *vi = EROFS_I(inode);
+
+	vi->fingerprint = *(struct erofs_inode_fingerprint *)data;
+	INIT_LIST_HEAD(&vi->ishare_list);
+	spin_lock_init(&vi->ishare_lock);
+	return 0;
+}
+
+bool erofs_ishare_fill_inode(struct inode *inode)
+{
+	static const struct file_operations empty_fops = {};
+	struct erofs_sb_info *sbi = EROFS_SB(inode->i_sb);
+	const struct address_space_operations *aops;
+	struct erofs_inode *vi = EROFS_I(inode);
+	struct erofs_inode_fingerprint fp;
+	struct dentry *sd;
+	struct inode *si;
+
+	aops = erofs_get_aops(inode);
+	if (IS_ERR(aops))
+		return false;
+	if (erofs_xattr_fill_inode_fingerprint(&fp, inode, sbi->domain_id))
+		return false;
+
+	si = iget5_locked(erofs_ishare_mnt->mnt_sb,
+			  xxh32(fp.opaque, fp.size, 0),
+			  erofs_ishare_iget5_eq, erofs_ishare_iget5_set, &fp);
+	if (si && (inode_state_read_once(si) & I_NEW)) {
+		si->i_fop = &empty_fops;
+		si->i_mapping->a_ops = aops;
+		si->i_mode = 0444 | S_IFREG;
+		si->i_size = inode->i_size;
+		unlock_new_inode(si);
+	} else {
+		kfree(fp.opaque);
+		if (!si || aops != si->i_mapping->a_ops) {
+			iput(si);
+			return false;
+		}
+		if (si->i_size != inode->i_size) {
+			erofs_warn(inode->i_sb, "i_size mismatch (%lld != %lld) for the same fingerprint",
+				   inode->i_size, si->i_size);
+			iput(si);
+			return false;
+		}
+	}
+	sd = d_obtain_alias(si); /* disconnected denties for sharedinodes */
+	if (IS_ERR(sd))
+		return false;
+	vi->sharedentry = sd;
+	INIT_LIST_HEAD(&vi->ishare_list);
+	spin_lock(&EROFS_I(si)->ishare_lock);
+	list_add(&vi->ishare_list, &EROFS_I(si)->ishare_list);
+	spin_unlock(&EROFS_I(si)->ishare_lock);
+	return true;
+}
+
+void erofs_ishare_free_inode(struct inode *inode)
+{
+	struct erofs_inode *vi = EROFS_I(inode), *svi;
+
+	if (!vi->sharedentry)
+		return;
+	svi = EROFS_I(d_inode(vi->sharedentry));
+	spin_lock(&svi->ishare_lock);
+	list_del(&vi->ishare_list);
+	spin_unlock(&svi->ishare_lock);
+	dput(vi->sharedentry);
+	vi->sharedentry = NULL;
+}
+
+static int erofs_ishare_file_open(struct inode *inode, struct file *file)
+{
+	struct path sharedpath = {
+		.mnt = erofs_ishare_mnt,
+		.dentry = EROFS_I(inode)->sharedentry,
+	};
+	struct file *rf;
+
+	if (file->f_flags & O_DIRECT)
+		return -EINVAL;
+
+	rf = backing_file_open(file, file->f_flags | O_NOATIME,
+			       &sharedpath, current_cred());
+	if (IS_ERR(rf))
+		return PTR_ERR(rf);
+	file->private_data = rf;
+	return 0;
+}
+
+static int erofs_ishare_file_release(struct inode *inode, struct file *file)
+{
+	fput(file->private_data);
+	file->private_data = NULL;
+	return 0;
+}
+
+static ssize_t erofs_ishare_file_read_iter(struct kiocb *iocb,
+					   struct iov_iter *to)
+{
+	struct file *realfile = iocb->ki_filp->private_data;
+	struct kiocb dedup_iocb;
+	ssize_t nread;
+
+	if (!iov_iter_count(to))
+		return 0;
+	kiocb_clone(&dedup_iocb, iocb, realfile);
+	nread = filemap_read(&dedup_iocb, to, 0);
+	iocb->ki_pos = dedup_iocb.ki_pos;
+	return nread;
+}
+
+static int erofs_ishare_mmap(struct file *file, struct vm_area_struct *vma)
+{
+	struct file *realfile = file->private_data;
+	int err;
+
+	vma_set_file(vma, realfile);
+
+	err = security_mmap_backing_file(vma, realfile, file);
+	if (err)
+		return err;
+
+	return generic_file_readonly_mmap(file, vma);
+}
+
+static int erofs_ishare_fadvise(struct file *file, loff_t offset,
+				loff_t len, int advice)
+{
+	return vfs_fadvise(file->private_data, offset, len, advice);
+}
+
+const struct file_operations erofs_ishare_fops = {
+	.open		= erofs_ishare_file_open,
+	.llseek		= generic_file_llseek,
+	.read_iter	= erofs_ishare_file_read_iter,
+	.mmap		= erofs_ishare_mmap,
+	.release	= erofs_ishare_file_release,
+	.get_unmapped_area = thp_get_unmapped_area,
+	.splice_read	= filemap_splice_read,
+	.fadvise	= erofs_ishare_fadvise,
+};
+
+struct inode *erofs_real_inode(struct inode *inode, bool *need_iput)
+{
+	struct erofs_inode *vi, *vi_share;
+	struct inode *realinode;
+
+	*need_iput = false;
+	if (inode->i_sb != erofs_ishare_mnt->mnt_sb)
+		return inode;
+
+	vi_share = EROFS_I(inode);
+	spin_lock(&vi_share->ishare_lock);
+	/* fetch any one as real inode */
+	DBG_BUGON(list_empty(&vi_share->ishare_list));
+	list_for_each_entry(vi, &vi_share->ishare_list, ishare_list) {
+		realinode = igrab(&vi->vfs_inode);
+		if (realinode) {
+			*need_iput = true;
+			break;
+		}
+	}
+	spin_unlock(&vi_share->ishare_lock);
+
+	DBG_BUGON(!realinode);
+	return realinode;
+}
+
+int __init erofs_init_ishare(void)
+{
+	struct vfsmount *mnt;
+	int ret;
+
+	mnt = kern_mount(&erofs_anon_fs_type);
+	if (IS_ERR(mnt))
+		return PTR_ERR(mnt);
+	/* generic_fadvise() doesn't work if s_bdi == &noop_backing_dev_info */
+	ret = super_setup_bdi(mnt->mnt_sb);
+	if (ret)
+		kern_unmount(mnt);
+	else
+		erofs_ishare_mnt = mnt;
+	return ret;
+}
+
+void erofs_exit_ishare(void)
+{
+	kern_unmount(erofs_ishare_mnt);
+}
diff --git a/fs/erofs/namei.c b/fs/erofs/namei.c
new file mode 100644
index 000000000..f7cf4f41a
--- /dev/null
+++ b/fs/erofs/namei.c
@@ -0,0 +1,224 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2017-2018 HUAWEI, Inc.
+ *             https://www.huawei.com/
+ * Copyright (C) 2022, Alibaba Cloud
+ */
+#include "xattr.h"
+#include <trace/events/erofs.h>
+
+struct erofs_qstr {
+	const unsigned char *name;
+	const unsigned char *end;
+};
+
+/* based on the end of qn is accurate and it must have the trailing '\0' */
+static inline int erofs_dirnamecmp(const struct erofs_qstr *qn,
+				   const struct erofs_qstr *qd,
+				   unsigned int *matched)
+{
+	unsigned int i = *matched;
+
+	/*
+	 * on-disk error, let's only BUG_ON in the debugging mode.
+	 * otherwise, it will return 1 to just skip the invalid name
+	 * and go on (in consideration of the lookup performance).
+	 */
+	DBG_BUGON(qd->name > qd->end);
+
+	/* qd could not have trailing '\0' */
+	/* However it is absolutely safe if < qd->end */
+	while (qd->name + i < qd->end && qd->name[i] != '\0') {
+		if (qn->name[i] != qd->name[i]) {
+			*matched = i;
+			return qn->name[i] > qd->name[i] ? 1 : -1;
+		}
+		++i;
+	}
+	*matched = i;
+	/* See comments in __d_alloc on the terminating NUL character */
+	return qn->name[i] == '\0' ? 0 : 1;
+}
+
+#define nameoff_from_disk(off, sz)	(le16_to_cpu(off) & ((sz) - 1))
+
+static struct erofs_dirent *find_target_dirent(struct erofs_qstr *name,
+					       u8 *data,
+					       unsigned int dirblksize,
+					       const int ndirents)
+{
+	int head, back;
+	unsigned int startprfx, endprfx;
+	struct erofs_dirent *const de = (struct erofs_dirent *)data;
+
+	/* since the 1st dirent has been evaluated previously */
+	head = 1;
+	back = ndirents - 1;
+	startprfx = endprfx = 0;
+
+	while (head <= back) {
+		const int mid = head + (back - head) / 2;
+		const int nameoff = nameoff_from_disk(de[mid].nameoff,
+						      dirblksize);
+		unsigned int matched = min(startprfx, endprfx);
+		struct erofs_qstr dname = {
+			.name = data + nameoff,
+			.end = mid >= ndirents - 1 ?
+				data + dirblksize :
+				data + nameoff_from_disk(de[mid + 1].nameoff,
+							 dirblksize)
+		};
+
+		/* string comparison without already matched prefix */
+		int ret = erofs_dirnamecmp(name, &dname, &matched);
+
+		if (!ret) {
+			return de + mid;
+		} else if (ret > 0) {
+			head = mid + 1;
+			startprfx = matched;
+		} else {
+			back = mid - 1;
+			endprfx = matched;
+		}
+	}
+
+	return ERR_PTR(-ENOENT);
+}
+
+static void *erofs_find_target_block(struct erofs_buf *target,
+		struct inode *dir, struct erofs_qstr *name, int *_ndirents)
+{
+	unsigned int bsz = i_blocksize(dir);
+	int head = 0, back = erofs_iblks(dir) - 1;
+	unsigned int startprfx = 0, endprfx = 0;
+	void *candidate = ERR_PTR(-ENOENT);
+
+	while (head <= back) {
+		const int mid = head + (back - head) / 2;
+		struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
+		struct erofs_dirent *de;
+
+		buf.mapping = dir->i_mapping;
+		de = erofs_bread(&buf, erofs_pos(dir->i_sb, mid), true);
+		if (!IS_ERR(de)) {
+			const int nameoff = nameoff_from_disk(de->nameoff, bsz);
+			const int ndirents = nameoff / sizeof(*de);
+			int diff;
+			unsigned int matched;
+			struct erofs_qstr dname;
+
+			if (!ndirents) {
+				erofs_put_metabuf(&buf);
+				erofs_err(dir->i_sb,
+					  "corrupted dir block %d @ nid %llu",
+					  mid, EROFS_I(dir)->nid);
+				DBG_BUGON(1);
+				de = ERR_PTR(-EFSCORRUPTED);
+				goto out;
+			}
+
+			matched = min(startprfx, endprfx);
+
+			dname.name = (u8 *)de + nameoff;
+			if (ndirents == 1)
+				dname.end = (u8 *)de + bsz;
+			else
+				dname.end = (u8 *)de +
+					nameoff_from_disk(de[1].nameoff, bsz);
+
+			/* string comparison without already matched prefix */
+			diff = erofs_dirnamecmp(name, &dname, &matched);
+
+			if (diff < 0) {
+				erofs_put_metabuf(&buf);
+				back = mid - 1;
+				endprfx = matched;
+				continue;
+			}
+
+			if (!IS_ERR(candidate))
+				erofs_put_metabuf(target);
+			*target = buf;
+			if (!diff) {
+				*_ndirents = 0;
+				return de;
+			}
+			head = mid + 1;
+			startprfx = matched;
+			candidate = de;
+			*_ndirents = ndirents;
+			continue;
+		}
+out:		/* free if the candidate is valid */
+		if (!IS_ERR(candidate))
+			erofs_put_metabuf(target);
+		return de;
+	}
+	return candidate;
+}
+
+int erofs_namei(struct inode *dir, const struct qstr *name, erofs_nid_t *nid,
+		unsigned int *d_type)
+{
+	int ndirents;
+	struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
+	struct erofs_dirent *de;
+	struct erofs_qstr qn;
+
+	if (!dir->i_size)
+		return -ENOENT;
+
+	qn.name = name->name;
+	qn.end = name->name + name->len;
+	buf.mapping = dir->i_mapping;
+
+	ndirents = 0;
+	de = erofs_find_target_block(&buf, dir, &qn, &ndirents);
+	if (IS_ERR(de))
+		return PTR_ERR(de);
+
+	if (ndirents)
+		de = find_target_dirent(&qn, (u8 *)de, i_blocksize(dir),
+					ndirents);
+
+	if (!IS_ERR(de)) {
+		*nid = le64_to_cpu(de->nid);
+		*d_type = de->file_type;
+	}
+	erofs_put_metabuf(&buf);
+	return PTR_ERR_OR_ZERO(de);
+}
+
+static struct dentry *erofs_lookup(struct inode *dir, struct dentry *dentry,
+				   unsigned int flags)
+{
+	int err;
+	erofs_nid_t nid;
+	unsigned int d_type;
+	struct inode *inode;
+
+	trace_erofs_lookup(dir, dentry, flags);
+
+	if (dentry->d_name.len > EROFS_NAME_LEN)
+		return ERR_PTR(-ENAMETOOLONG);
+
+	err = erofs_namei(dir, &dentry->d_name, &nid, &d_type);
+
+	if (err == -ENOENT)
+		/* negative dentry */
+		inode = NULL;
+	else if (err)
+		inode = ERR_PTR(err);
+	else
+		inode = erofs_iget(dir->i_sb, nid);
+	return d_splice_alias(inode, dentry);
+}
+
+const struct inode_operations erofs_dir_iops = {
+	.lookup = erofs_lookup,
+	.getattr = erofs_getattr,
+	.listxattr = erofs_listxattr,
+	.get_inode_acl = erofs_get_acl,
+	.fiemap = erofs_fiemap,
+};
diff --git a/fs/erofs/super.c b/fs/erofs/super.c
new file mode 100644
index 000000000..0967aace9
--- /dev/null
+++ b/fs/erofs/super.c
@@ -0,0 +1,1146 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2017-2018 HUAWEI, Inc.
+ *             https://www.huawei.com/
+ * Copyright (C) 2021, Alibaba Cloud
+ */
+#include <linux/statfs.h>
+#include <linux/seq_file.h>
+#include <linux/crc32c.h>
+#include <linux/fs_context.h>
+#include <linux/fs_parser.h>
+#include <linux/exportfs.h>
+#include <linux/backing-dev.h>
+#include <linux/pseudo_fs.h>
+#include "xattr.h"
+
+#define CREATE_TRACE_POINTS
+#include <trace/events/erofs.h>
+
+static struct kmem_cache *erofs_inode_cachep __read_mostly;
+
+void _erofs_printk(struct super_block *sb, const char *fmt, ...)
+{
+	struct va_format vaf;
+	va_list args;
+	int level;
+
+	va_start(args, fmt);
+
+	level = printk_get_level(fmt);
+	vaf.fmt = printk_skip_level(fmt);
+	vaf.va = &args;
+	if (sb)
+		printk("%c%cerofs (device %s): %pV",
+				KERN_SOH_ASCII, level, sb->s_id, &vaf);
+	else
+		printk("%c%cerofs: %pV", KERN_SOH_ASCII, level, &vaf);
+	va_end(args);
+}
+
+static int erofs_superblock_csum_verify(struct super_block *sb, void *sbdata)
+{
+	struct erofs_super_block *dsb = sbdata + EROFS_SUPER_OFFSET;
+	u32 len = 1 << EROFS_SB(sb)->blkszbits, crc;
+
+	if (len > EROFS_SUPER_OFFSET)
+		len -= EROFS_SUPER_OFFSET;
+	len -= offsetof(struct erofs_super_block, checksum) +
+			sizeof(dsb->checksum);
+
+	/* skip .magic(pre-verified) and .checksum(0) fields */
+	crc = crc32c(0x5045B54A, (&dsb->checksum) + 1, len);
+	if (crc == le32_to_cpu(dsb->checksum))
+		return 0;
+	erofs_err(sb, "invalid checksum 0x%08x, 0x%08x expected",
+		  crc, le32_to_cpu(dsb->checksum));
+	return -EBADMSG;
+}
+
+static void erofs_inode_init_once(void *ptr)
+{
+	struct erofs_inode *vi = ptr;
+
+	inode_init_once(&vi->vfs_inode);
+}
+
+static struct inode *erofs_alloc_inode(struct super_block *sb)
+{
+	struct erofs_inode *vi =
+		alloc_inode_sb(sb, erofs_inode_cachep, GFP_KERNEL);
+
+	if (!vi)
+		return NULL;
+
+	/* zero out everything except vfs_inode */
+	memset(vi, 0, offsetof(struct erofs_inode, vfs_inode));
+	return &vi->vfs_inode;
+}
+
+static void erofs_free_inode(struct inode *inode)
+{
+	struct erofs_inode *vi = EROFS_I(inode);
+
+	if (inode->i_op == &erofs_fast_symlink_iops)
+		kfree(inode->i_link);
+	kfree(vi->xattr_shared_xattrs);
+	kmem_cache_free(erofs_inode_cachep, vi);
+}
+
+/* read variable-sized metadata, offset will be aligned by 4-byte */
+void *erofs_read_metadata(struct super_block *sb, struct erofs_buf *buf,
+			  erofs_off_t *offset, int *lengthp)
+{
+	u8 *buffer, *ptr;
+	int len, i, cnt;
+
+	*offset = round_up(*offset, 4);
+	ptr = erofs_bread(buf, *offset, true);
+	if (IS_ERR(ptr))
+		return ptr;
+
+	len = le16_to_cpu(*(__le16 *)ptr);
+	if (!len)
+		len = U16_MAX + 1;
+	buffer = kmalloc(len, GFP_KERNEL);
+	if (!buffer)
+		return ERR_PTR(-ENOMEM);
+	*offset += sizeof(__le16);
+	*lengthp = len;
+
+	for (i = 0; i < len; i += cnt) {
+		cnt = min_t(int, sb->s_blocksize - erofs_blkoff(sb, *offset),
+			    len - i);
+		ptr = erofs_bread(buf, *offset, true);
+		if (IS_ERR(ptr)) {
+			kfree(buffer);
+			return ptr;
+		}
+		memcpy(buffer + i, ptr, cnt);
+		*offset += cnt;
+	}
+	return buffer;
+}
+
+static int erofs_init_device(struct erofs_buf *buf, struct super_block *sb,
+			     struct erofs_device_info *dif, erofs_off_t *pos)
+{
+	struct erofs_sb_info *sbi = EROFS_SB(sb);
+	struct erofs_deviceslot *dis;
+	struct file *file;
+	bool _48bit;
+
+	dis = erofs_read_metabuf(buf, sb, *pos, false);
+	if (IS_ERR(dis))
+		return PTR_ERR(dis);
+
+	if (!sbi->devs->flatdev && !dif->path) {
+		if (!dis->tag[0]) {
+			erofs_err(sb, "empty device tag @ pos %llu", *pos);
+			return -EINVAL;
+		}
+		dif->path = kmemdup_nul(dis->tag, sizeof(dis->tag), GFP_KERNEL);
+		if (!dif->path)
+			return -ENOMEM;
+	}
+
+	if (!sbi->devs->flatdev) {
+		file = erofs_is_fileio_mode(sbi) ?
+				filp_open(dif->path, O_RDONLY | O_LARGEFILE, 0) :
+				fs_bdev_file_open_by_path(dif->path,
+						BLK_OPEN_READ, sb->s_type, sb);
+		if (IS_ERR(file)) {
+			if (file == ERR_PTR(-ENOTBLK))
+				return -EINVAL;
+			return PTR_ERR(file);
+		}
+
+		if (!erofs_is_fileio_mode(sbi)) {
+			dif->dax_dev = fs_dax_get_by_bdev(file_bdev(file),
+					&dif->dax_part_off, NULL, NULL);
+		} else if (!S_ISREG(file_inode(file)->i_mode)) {
+			fput(file);
+			return -EINVAL;
+		}
+		if (!dif->dax_dev && test_opt(&sbi->opt, DAX_ALWAYS)) {
+			erofs_info(sb, "DAX unsupported by %s. Turning off DAX.",
+				   dif->path);
+			clear_opt(&sbi->opt, DAX_ALWAYS);
+		}
+		dif->file = file;
+	}
+
+	_48bit = erofs_sb_has_48bit(sbi);
+	dif->blocks = le32_to_cpu(dis->blocks_lo) |
+		(_48bit ? (u64)le16_to_cpu(dis->blocks_hi) << 32 : 0);
+	dif->uniaddr = le32_to_cpu(dis->uniaddr_lo) |
+		(_48bit ? (u64)le16_to_cpu(dis->uniaddr_hi) << 32 : 0);
+	sbi->total_blocks += dif->blocks;
+	*pos += EROFS_DEVT_SLOT_SIZE;
+	return 0;
+}
+
+static int erofs_scan_devices(struct super_block *sb,
+			      struct erofs_super_block *dsb)
+{
+	struct erofs_sb_info *sbi = EROFS_SB(sb);
+	unsigned int ondisk_extradevs;
+	erofs_off_t pos;
+	struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
+	struct erofs_device_info *dif;
+	int id, err = 0;
+
+	sbi->total_blocks = sbi->dif0.blocks;
+	if (!erofs_sb_has_device_table(sbi))
+		ondisk_extradevs = 0;
+	else
+		ondisk_extradevs = le16_to_cpu(dsb->extra_devices);
+
+	if (sbi->devs->extra_devices &&
+	    ondisk_extradevs != sbi->devs->extra_devices) {
+		erofs_err(sb, "extra devices don't match (ondisk %u, given %u)",
+			  ondisk_extradevs, sbi->devs->extra_devices);
+		return -EINVAL;
+	}
+
+	if (test_opt(&sbi->opt, DAX_ALWAYS) && !sbi->dif0.dax_dev) {
+		erofs_info(sb, "DAX unsupported by block device. Turning off DAX.");
+		clear_opt(&sbi->opt, DAX_ALWAYS);
+	}
+	if (!ondisk_extradevs)
+		return 0;
+
+	if (!sbi->devs->extra_devices)
+		sbi->devs->flatdev = true;
+
+	sbi->device_id_mask = roundup_pow_of_two(ondisk_extradevs + 1) - 1;
+	pos = le16_to_cpu(dsb->devt_slotoff) * EROFS_DEVT_SLOT_SIZE;
+	down_read(&sbi->devs->rwsem);
+	if (sbi->devs->extra_devices) {
+		idr_for_each_entry(&sbi->devs->tree, dif, id) {
+			err = erofs_init_device(&buf, sb, dif, &pos);
+			if (err)
+				break;
+		}
+	} else {
+		for (id = 0; id < ondisk_extradevs; id++) {
+			dif = kzalloc_obj(*dif);
+			if (!dif) {
+				err = -ENOMEM;
+				break;
+			}
+
+			err = idr_alloc(&sbi->devs->tree, dif, 0, 0, GFP_KERNEL);
+			if (err < 0) {
+				kfree(dif);
+				break;
+			}
+			++sbi->devs->extra_devices;
+
+			err = erofs_init_device(&buf, sb, dif, &pos);
+			if (err)
+				break;
+		}
+	}
+	up_read(&sbi->devs->rwsem);
+	erofs_put_metabuf(&buf);
+	return err;
+}
+
+static int erofs_read_superblock(struct super_block *sb)
+{
+	struct erofs_sb_info *sbi = EROFS_SB(sb);
+	struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
+	struct erofs_super_block *dsb;
+	void *data;
+	int ret;
+
+	data = erofs_read_metabuf(&buf, sb, 0, false);
+	if (IS_ERR(data)) {
+		erofs_err(sb, "cannot read erofs superblock");
+		return PTR_ERR(data);
+	}
+
+	dsb = (struct erofs_super_block *)(data + EROFS_SUPER_OFFSET);
+	ret = -EINVAL;
+	if (le32_to_cpu(dsb->magic) != EROFS_SUPER_MAGIC_V1) {
+		erofs_err(sb, "cannot find valid erofs superblock");
+		goto out;
+	}
+
+	sbi->blkszbits = dsb->blkszbits;
+	if (sbi->blkszbits < 9 || sbi->blkszbits > PAGE_SHIFT) {
+		erofs_err(sb, "blkszbits %u isn't supported", sbi->blkszbits);
+		goto out;
+	}
+	if (dsb->dirblkbits) {
+		erofs_err(sb, "dirblkbits %u isn't supported", dsb->dirblkbits);
+		goto out;
+	}
+
+	sbi->feature_compat = le32_to_cpu(dsb->feature_compat);
+	if (erofs_sb_has_sb_chksum(sbi)) {
+		ret = erofs_superblock_csum_verify(sb, data);
+		if (ret)
+			goto out;
+	}
+
+	ret = -EINVAL;
+	sbi->feature_incompat = le32_to_cpu(dsb->feature_incompat);
+	if (sbi->feature_incompat & ~EROFS_ALL_FEATURE_INCOMPAT) {
+		erofs_err(sb, "unidentified incompatible feature %x, please upgrade kernel",
+			  sbi->feature_incompat & ~EROFS_ALL_FEATURE_INCOMPAT);
+		goto out;
+	}
+
+	sbi->sb_size = 128 + dsb->sb_extslots * EROFS_SB_EXTSLOT_SIZE;
+	if (sbi->sb_size > PAGE_SIZE - EROFS_SUPER_OFFSET) {
+		erofs_err(sb, "invalid sb_extslots %u (more than a fs block)",
+			  sbi->sb_size);
+		goto out;
+	}
+	sbi->dif0.blocks = le32_to_cpu(dsb->blocks_lo);
+	sbi->meta_blkaddr = le32_to_cpu(dsb->meta_blkaddr);
+#ifdef CONFIG_EROFS_FS_XATTR
+	sbi->xattr_blkaddr = le32_to_cpu(dsb->xattr_blkaddr);
+	sbi->xattr_prefix_start = le32_to_cpu(dsb->xattr_prefix_start);
+	sbi->xattr_prefix_count = dsb->xattr_prefix_count;
+	sbi->xattr_filter_reserved = dsb->xattr_filter_reserved;
+	if (erofs_sb_has_ishare_xattrs(sbi)) {
+		if (dsb->ishare_xattr_prefix_id >= sbi->xattr_prefix_count) {
+			erofs_err(sb, "invalid ishare xattr prefix id %u",
+				  dsb->ishare_xattr_prefix_id);
+			ret = -EFSCORRUPTED;
+			goto out;
+		}
+		sbi->ishare_xattr_prefix_id = dsb->ishare_xattr_prefix_id;
+	}
+#endif
+	sbi->islotbits = ilog2(sizeof(struct erofs_inode_compact));
+	if (erofs_sb_has_48bit(sbi) && dsb->rootnid_8b) {
+		sbi->root_nid = le64_to_cpu(dsb->rootnid_8b);
+		sbi->dif0.blocks = sbi->dif0.blocks |
+				((u64)le16_to_cpu(dsb->rb.blocks_hi) << 32);
+	} else {
+		sbi->root_nid = le16_to_cpu(dsb->rb.rootnid_2b);
+	}
+	sbi->packed_nid = le64_to_cpu(dsb->packed_nid);
+	if (erofs_sb_has_metabox(sbi)) {
+		ret = -EFSCORRUPTED;
+		if (sbi->sb_size <= offsetof(struct erofs_super_block,
+					     metabox_nid))
+			goto out;
+		sbi->metabox_nid = le64_to_cpu(dsb->metabox_nid);
+		if (sbi->metabox_nid & BIT_ULL(EROFS_DIRENT_NID_METABOX_BIT))
+			goto out;		/* self-loop detection */
+	}
+	sbi->inos = le64_to_cpu(dsb->inos);
+
+	sbi->epoch = (s64)le64_to_cpu(dsb->epoch);
+	sbi->fixed_nsec = le32_to_cpu(dsb->fixed_nsec);
+	super_set_uuid(sb, (void *)dsb->uuid, sizeof(dsb->uuid));
+
+	if (dsb->volume_name[0]) {
+		sbi->volume_name = kstrndup(dsb->volume_name,
+					    sizeof(dsb->volume_name), GFP_KERNEL);
+		if (!sbi->volume_name) {
+			ret = -ENOMEM;
+			goto out;
+		}
+	}
+
+	if (IS_ENABLED(CONFIG_EROFS_FS_ZIP)) {
+		ret = z_erofs_parse_cfgs(sb, dsb);
+		if (ret < 0)
+			goto out;
+	} else if (dsb->u1.available_compr_algs ||
+		   erofs_sb_has_lz4_0padding(sbi)) {
+		erofs_err(sb, "compression disabled, unable to mount compressed EROFS");
+		ret = -EOPNOTSUPP;
+		goto out;
+	}
+
+	ret = erofs_scan_devices(sb, dsb);
+
+	if (erofs_sb_has_48bit(sbi))
+		erofs_info(sb, "EXPERIMENTAL 48-bit layout support in use. Use at your own risk!");
+	if (erofs_sb_has_metabox(sbi))
+		erofs_info(sb, "EXPERIMENTAL metadata compression support in use. Use at your own risk!");
+out:
+	erofs_put_metabuf(&buf);
+	return ret;
+}
+
+static void erofs_default_options(struct erofs_sb_info *sbi)
+{
+#ifdef CONFIG_EROFS_FS_ZIP
+	sbi->opt.cache_strategy = EROFS_ZIP_CACHE_READAROUND;
+	sbi->sync_decompress = EROFS_SYNC_DECOMPRESS_AUTO;
+#endif
+	if (IS_ENABLED(CONFIG_EROFS_FS_XATTR))
+		set_opt(&sbi->opt, XATTR_USER);
+	if (IS_ENABLED(CONFIG_EROFS_FS_POSIX_ACL))
+		set_opt(&sbi->opt, POSIX_ACL);
+}
+
+enum {
+	Opt_user_xattr, Opt_acl, Opt_cache_strategy, Opt_dax, Opt_dax_enum,
+	Opt_device, Opt_domain_id, Opt_directio, Opt_fsoffset, Opt_inode_share,
+	Opt_source,
+};
+
+static const struct constant_table erofs_param_cache_strategy[] = {
+	{"disabled",	EROFS_ZIP_CACHE_DISABLED},
+	{"readahead",	EROFS_ZIP_CACHE_READAHEAD},
+	{"readaround",	EROFS_ZIP_CACHE_READAROUND},
+	{}
+};
+
+static const struct constant_table erofs_dax_param_enums[] = {
+	{"always",	EROFS_MOUNT_DAX_ALWAYS},
+	{"never",	EROFS_MOUNT_DAX_NEVER},
+	{}
+};
+
+static const struct fs_parameter_spec erofs_fs_parameters[] = {
+	fsparam_flag_no("user_xattr",		Opt_user_xattr),
+	fsparam_flag_no("acl",			Opt_acl),
+	fsparam_enum("cache_strategy",		Opt_cache_strategy,
+		     erofs_param_cache_strategy),
+	fsparam_flag("dax",			Opt_dax),
+	fsparam_enum("dax",			Opt_dax_enum, erofs_dax_param_enums),
+	fsparam_string("device",		Opt_device),
+	fsparam_string("domain_id",		Opt_domain_id),
+	fsparam_flag_no("directio",		Opt_directio),
+	fsparam_u64("fsoffset",			Opt_fsoffset),
+	fsparam_flag("inode_share",		Opt_inode_share),
+	fsparam_file_or_string("source",	Opt_source),
+	{}
+};
+
+static bool erofs_fc_set_dax_mode(struct fs_context *fc, unsigned int mode)
+{
+	if (IS_ENABLED(CONFIG_FS_DAX)) {
+		struct erofs_sb_info *sbi = fc->s_fs_info;
+
+		if (mode == EROFS_MOUNT_DAX_ALWAYS) {
+			set_opt(&sbi->opt, DAX_ALWAYS);
+			clear_opt(&sbi->opt, DAX_NEVER);
+			return true;
+		} else if (mode == EROFS_MOUNT_DAX_NEVER) {
+			set_opt(&sbi->opt, DAX_NEVER);
+			clear_opt(&sbi->opt, DAX_ALWAYS);
+			return true;
+		}
+		DBG_BUGON(1);
+		return false;
+	}
+	errorfc(fc, "dax options not supported");
+	return false;
+}
+
+static int erofs_fc_parse_source(struct fs_context *fc,
+				 struct fs_parameter *param)
+{
+	struct erofs_sb_info *sbi = fc->s_fs_info;
+
+	if (fc->source || sbi->dif0.file)
+		return invalf(fc, "Multiple sources");
+
+	switch (param->type) {
+	case fs_value_is_string:
+		fc->source = param->string;
+		param->string = NULL;
+		return 0;
+	case fs_value_is_file: {
+		char *buf __free(kfree) = kmalloc(PATH_MAX, GFP_KERNEL);
+		char *p;
+
+		if (!buf)
+			return -ENOMEM;
+		p = file_path(param->file, buf, PATH_MAX);
+		if (IS_ERR(p))
+			return PTR_ERR(p);
+		fc->source = kstrdup(p, GFP_KERNEL);
+		if (!fc->source)
+			return -ENOMEM;
+		sbi->dif0.file = no_free_ptr(param->file);
+		return 0;
+	}
+	default:
+		WARN_ON_ONCE(true);
+		return -EINVAL;
+	}
+}
+
+static int erofs_fc_parse_param(struct fs_context *fc,
+				struct fs_parameter *param)
+{
+	struct erofs_sb_info *sbi = fc->s_fs_info;
+	struct fs_parse_result result;
+	struct erofs_device_info *dif;
+	int opt, ret;
+
+	opt = fs_parse(fc, erofs_fs_parameters, param, &result);
+	if (opt < 0)
+		return opt;
+
+	switch (opt) {
+	case Opt_user_xattr:
+		if (!IS_ENABLED(CONFIG_EROFS_FS_XATTR))
+			errorfc(fc, "{,no}user_xattr options not supported");
+		else if (result.boolean)
+			set_opt(&sbi->opt, XATTR_USER);
+		else
+			clear_opt(&sbi->opt, XATTR_USER);
+		break;
+	case Opt_acl:
+		if (!IS_ENABLED(CONFIG_EROFS_FS_POSIX_ACL))
+			errorfc(fc, "{,no}acl options not supported");
+		else if (result.boolean)
+			set_opt(&sbi->opt, POSIX_ACL);
+		else
+			clear_opt(&sbi->opt, POSIX_ACL);
+		break;
+	case Opt_cache_strategy:
+		if (!IS_ENABLED(CONFIG_EROFS_FS_ZIP))
+			errorfc(fc, "compression not supported, cache_strategy ignored");
+		else
+			sbi->opt.cache_strategy = result.uint_32;
+		break;
+	case Opt_dax:
+		if (!erofs_fc_set_dax_mode(fc, EROFS_MOUNT_DAX_ALWAYS))
+			return -EINVAL;
+		break;
+	case Opt_dax_enum:
+		if (!erofs_fc_set_dax_mode(fc, result.uint_32))
+			return -EINVAL;
+		break;
+	case Opt_device:
+		dif = kzalloc_obj(*dif);
+		if (!dif)
+			return -ENOMEM;
+		dif->path = kstrdup(param->string, GFP_KERNEL);
+		if (!dif->path) {
+			kfree(dif);
+			return -ENOMEM;
+		}
+		down_write(&sbi->devs->rwsem);
+		ret = idr_alloc(&sbi->devs->tree, dif, 0, 0, GFP_KERNEL);
+		up_write(&sbi->devs->rwsem);
+		if (ret < 0) {
+			kfree(dif->path);
+			kfree(dif);
+			return ret;
+		}
+		++sbi->devs->extra_devices;
+		break;
+	case Opt_domain_id:
+		if (!IS_ENABLED(CONFIG_EROFS_FS_PAGE_CACHE_SHARE)) {
+			errorfc(fc, "%s option not supported", erofs_fs_parameters[opt].name);
+		} else {
+			kfree_sensitive(sbi->domain_id);
+			sbi->domain_id = no_free_ptr(param->string);
+		}
+		break;
+	case Opt_directio:
+		if (!IS_ENABLED(CONFIG_EROFS_FS_BACKED_BY_FILE))
+			errorfc(fc, "%s option not supported", erofs_fs_parameters[opt].name);
+		else if (result.boolean)
+			set_opt(&sbi->opt, DIRECT_IO);
+		else
+			clear_opt(&sbi->opt, DIRECT_IO);
+		break;
+	case Opt_fsoffset:
+		sbi->dif0.fsoff = result.uint_64;
+		break;
+	case Opt_inode_share:
+		if (!IS_ENABLED(CONFIG_EROFS_FS_PAGE_CACHE_SHARE))
+			errorfc(fc, "%s option not supported", erofs_fs_parameters[opt].name);
+		else
+			set_opt(&sbi->opt, INODE_SHARE);
+		break;
+	case Opt_source:
+		return erofs_fc_parse_source(fc, param);
+	}
+	return 0;
+}
+
+static int erofs_encode_fh(struct inode *inode, u32 *fh, int *max_len,
+			   struct inode *parent)
+{
+	erofs_nid_t nid = EROFS_I(inode)->nid;
+	int len = parent ? 6 : 3;
+
+	if (*max_len < len) {
+		*max_len = len;
+		return FILEID_INVALID;
+	}
+
+	fh[0] = (u32)(nid >> 32);
+	fh[1] = (u32)(nid & 0xffffffff);
+	fh[2] = inode->i_generation;
+
+	if (parent) {
+		nid = EROFS_I(parent)->nid;
+
+		fh[3] = (u32)(nid >> 32);
+		fh[4] = (u32)(nid & 0xffffffff);
+		fh[5] = parent->i_generation;
+	}
+
+	*max_len = len;
+	return parent ? FILEID_INO64_GEN_PARENT : FILEID_INO64_GEN;
+}
+
+static struct dentry *erofs_fh_to_dentry(struct super_block *sb,
+		struct fid *fid, int fh_len, int fh_type)
+{
+	if ((fh_type != FILEID_INO64_GEN &&
+	     fh_type != FILEID_INO64_GEN_PARENT) || fh_len < 3)
+		return NULL;
+
+	return d_obtain_alias(erofs_iget(sb,
+		((u64)fid->raw[0] << 32) | fid->raw[1]));
+}
+
+static struct dentry *erofs_fh_to_parent(struct super_block *sb,
+		struct fid *fid, int fh_len, int fh_type)
+{
+	if (fh_type != FILEID_INO64_GEN_PARENT || fh_len < 6)
+		return NULL;
+
+	return d_obtain_alias(erofs_iget(sb,
+		((u64)fid->raw[3] << 32) | fid->raw[4]));
+}
+
+static struct dentry *erofs_get_parent(struct dentry *child)
+{
+	erofs_nid_t nid;
+	unsigned int d_type;
+	int err;
+
+	err = erofs_namei(d_inode(child), &dotdot_name, &nid, &d_type);
+	if (err)
+		return ERR_PTR(err);
+	return d_obtain_alias(erofs_iget(child->d_sb, nid));
+}
+
+static const struct export_operations erofs_export_ops = {
+	.encode_fh = erofs_encode_fh,
+	.fh_to_dentry = erofs_fh_to_dentry,
+	.fh_to_parent = erofs_fh_to_parent,
+	.get_parent = erofs_get_parent,
+};
+
+int erofs_setup_managed_cache(struct super_block *sb)
+{
+	if (!EROFS_SB(sb)->managed_cache) {
+		struct inode *inode = new_inode(sb);
+
+		if (!inode)
+			return -ENOMEM;
+		set_nlink(inode, 1);
+		inode->i_size = OFFSET_MAX;
+		mapping_set_gfp_mask(inode->i_mapping, GFP_KERNEL);
+		EROFS_SB(sb)->managed_cache = inode;
+	}
+	return 0;
+}
+
+static int erofs_fc_fill_super(struct super_block *sb, struct fs_context *fc)
+{
+	struct inode *inode;
+	struct erofs_sb_info *sbi = EROFS_SB(sb);
+	int err;
+
+	sb->s_magic = EROFS_SUPER_MAGIC;
+	sb->s_flags |= SB_RDONLY | SB_NOATIME;
+	sb->s_maxbytes = MAX_LFS_FILESIZE;
+	sb->s_op = &erofs_sops;
+
+	if (!sbi->domain_id && test_opt(&sbi->opt, INODE_SHARE)) {
+		errorfc(fc, "domain_id is needed when inode_share is on");
+		return -EINVAL;
+	}
+	if (test_opt(&sbi->opt, DAX_ALWAYS) && test_opt(&sbi->opt, INODE_SHARE)) {
+		errorfc(fc, "FSDAX is not allowed when inode_share is on");
+		return -EINVAL;
+	}
+
+	sbi->blkszbits = PAGE_SHIFT;
+	if (erofs_is_fileio_mode(sbi)) {
+		/*
+		 * (File-backed mounts) EROFS claims it's safe to nest other
+		 * fs contexts (including its own) due to self-controlled RO
+		 * accesses/contexts and no side-effect changes that need to
+		 * context save & restore so it can reuse the current thread
+		 * context.
+		 * However, we still need to prevent kernel stack overflow due
+		 * to filesystem nesting: just ensure that s_stack_depth is 0
+		 * to disallow mounting EROFS on stacked filesystems.
+		 * Note: s_stack_depth is not incremented here for now, since
+		 * EROFS is the only fs supporting file-backed mounts for now.
+		 * It MUST change if another fs plans to support them, which
+		 * may also require adjusting FILESYSTEM_MAX_STACK_DEPTH.
+		 */
+		inode = file_inode(sbi->dif0.file);
+		if ((inode->i_sb->s_op == &erofs_sops &&
+		     !inode->i_sb->s_bdev) || inode->i_sb->s_stack_depth) {
+			erofs_err(sb, "file-backed mounts cannot be applied to stacked fses");
+			return -ENOTBLK;
+		}
+		sb->s_blocksize = PAGE_SIZE;
+		sb->s_blocksize_bits = PAGE_SHIFT;
+
+		err = super_setup_bdi(sb);
+		if (err)
+			return err;
+		err = erofs_setup_managed_cache(sb);
+		if (err)
+			return err;
+
+		snprintf(sb->s_id, sizeof(sb->s_id),
+			 "%u:%u", MAJOR(sb->s_dev), MINOR(sb->s_dev));
+	} else {
+		if (!sb_set_blocksize(sb, PAGE_SIZE)) {
+			errorfc(fc, "failed to set initial blksize");
+			return -EINVAL;
+		}
+		sbi->dif0.dax_dev = fs_dax_get_by_bdev(sb->s_bdev,
+				&sbi->dif0.dax_part_off, NULL, NULL);
+	}
+
+	err = erofs_read_superblock(sb);
+	if (err)
+		return err;
+
+	if (sb->s_blocksize_bits != sbi->blkszbits) {
+		if (erofs_is_fileio_mode(sbi)) {
+			sb->s_blocksize = 1 << sbi->blkszbits;
+			sb->s_blocksize_bits = sbi->blkszbits;
+		} else if (!sb_set_blocksize(sb, 1 << sbi->blkszbits)) {
+			errorfc(fc, "failed to set erofs blksize");
+			return -EINVAL;
+		}
+	}
+	if (sbi->dif0.fsoff & (sb->s_blocksize - 1))
+		return invalfc(fc, "fsoffset %llu is not aligned to block size %lu",
+			       sbi->dif0.fsoff, sb->s_blocksize);
+
+	if (test_opt(&sbi->opt, DAX_ALWAYS) && sbi->blkszbits != PAGE_SHIFT) {
+		erofs_info(sb, "unsupported blocksize for DAX");
+		clear_opt(&sbi->opt, DAX_ALWAYS);
+	}
+	if (test_opt(&sbi->opt, INODE_SHARE) && !erofs_sb_has_ishare_xattrs(sbi)) {
+		erofs_info(sb, "on-disk ishare xattrs not found. Turning off inode_share.");
+		clear_opt(&sbi->opt, INODE_SHARE);
+	}
+	if (test_opt(&sbi->opt, INODE_SHARE))
+		erofs_info(sb, "EXPERIMENTAL EROFS page cache share support in use. Use at your own risk!");
+
+	sb->s_time_gran = 1;
+	sb->s_xattr = erofs_xattr_handlers;
+	sb->s_export_op = &erofs_export_ops;
+
+	if (test_opt(&sbi->opt, POSIX_ACL))
+		sb->s_flags |= SB_POSIXACL;
+	else
+		sb->s_flags &= ~SB_POSIXACL;
+
+	err = z_erofs_init_super(sb);
+	if (err)
+		return err;
+
+	if (erofs_sb_has_fragments(sbi) && sbi->packed_nid) {
+		inode = erofs_iget(sb, sbi->packed_nid);
+		if (IS_ERR(inode))
+			return PTR_ERR(inode);
+		sbi->packed_inode = inode;
+	}
+	if (erofs_sb_has_metabox(sbi)) {
+		inode = erofs_iget(sb, sbi->metabox_nid);
+		if (IS_ERR(inode))
+			return PTR_ERR(inode);
+		sbi->metabox_inode = inode;
+	}
+
+	inode = erofs_iget(sb, sbi->root_nid);
+	if (IS_ERR(inode))
+		return PTR_ERR(inode);
+
+	if (!S_ISDIR(inode->i_mode)) {
+		erofs_err(sb, "rootino(nid %llu) is not a directory(i_mode %o)",
+			  sbi->root_nid, inode->i_mode);
+		iput(inode);
+		return -EINVAL;
+	}
+	sb->s_root = d_make_root(inode);
+	if (!sb->s_root)
+		return -ENOMEM;
+
+	erofs_shrinker_register(sb);
+	err = erofs_xattr_prefixes_init(sb);
+	if (err)
+		return err;
+
+	super_set_sysfs_name_id(sb);
+	err = erofs_register_sysfs(sb);
+	if (err)
+		return err;
+
+	sbi->dir_ra_bytes = EROFS_DIR_RA_BYTES;
+	erofs_info(sb, "mounted with root inode @ nid %llu.", sbi->root_nid);
+	return 0;
+}
+
+static int erofs_fc_get_tree(struct fs_context *fc)
+{
+	struct erofs_sb_info *sbi = fc->s_fs_info;
+	int ret;
+
+	if (sbi->dif0.file) {
+		if (!IS_ENABLED(CONFIG_EROFS_FS_BACKED_BY_FILE)) {
+			errorfc(fc, "source fd option not supported");
+			return -EINVAL;
+		}
+		if (!S_ISREG(file_inode(sbi->dif0.file)->i_mode) ||
+		    !sbi->dif0.file->f_mapping->a_ops->read_folio) {
+			errorfc(fc, "source is unsupported");
+			return -EINVAL;
+		}
+		return get_tree_nodev(fc, erofs_fc_fill_super);
+	}
+
+	ret = get_tree_bdev_flags(fc, erofs_fc_fill_super,
+		IS_ENABLED(CONFIG_EROFS_FS_BACKED_BY_FILE) ?
+			GET_TREE_BDEV_QUIET_LOOKUP : 0);
+	if (IS_ENABLED(CONFIG_EROFS_FS_BACKED_BY_FILE) && ret == -ENOTBLK) {
+		struct file *file;
+
+		if (!fc->source)
+			return invalf(fc, "No source specified");
+		file = filp_open(fc->source, O_RDONLY | O_LARGEFILE, 0);
+		if (IS_ERR(file))
+			return PTR_ERR(file);
+		sbi->dif0.file = file;
+
+		if (S_ISREG(file_inode(sbi->dif0.file)->i_mode) &&
+		    sbi->dif0.file->f_mapping->a_ops->read_folio)
+			return get_tree_nodev(fc, erofs_fc_fill_super);
+	}
+	return ret;
+}
+
+static int erofs_fc_reconfigure(struct fs_context *fc)
+{
+	struct super_block *sb = fc->root->d_sb;
+	struct erofs_sb_info *sbi = EROFS_SB(sb);
+	struct erofs_sb_info *new_sbi = fc->s_fs_info;
+
+	DBG_BUGON(!sb_rdonly(sb));
+
+	if (new_sbi->domain_id)
+		erofs_info(sb, "ignoring reconfiguration for domain_id.");
+
+	if (test_opt(&new_sbi->opt, POSIX_ACL))
+		fc->sb_flags |= SB_POSIXACL;
+	else
+		fc->sb_flags &= ~SB_POSIXACL;
+
+	sbi->opt = new_sbi->opt;
+
+	fc->sb_flags |= SB_RDONLY;
+	return 0;
+}
+
+static int erofs_release_device_info(int id, void *ptr, void *data)
+{
+	struct super_block *sb = data;
+	struct erofs_device_info *dif = ptr;
+
+	fs_put_dax(dif->dax_dev, NULL);
+	if (dif->file) {
+		if (S_ISBLK(file_inode(dif->file)->i_mode))
+			fs_bdev_file_release(dif->file, sb);
+		else
+			fput(dif->file);
+	}
+	kfree(dif->path);
+	kfree(dif);
+	return 0;
+}
+
+static void erofs_free_dev_context(struct erofs_dev_context *devs,
+				   struct super_block *sb)
+{
+	if (!devs)
+		return;
+	idr_for_each(&devs->tree, &erofs_release_device_info, sb);
+	idr_destroy(&devs->tree);
+	kfree(devs);
+}
+
+static void erofs_sb_free(struct erofs_sb_info *sbi, struct super_block *sb)
+{
+	erofs_free_dev_context(sbi->devs, sb);
+	kfree_sensitive(sbi->domain_id);
+	if (sbi->dif0.file)
+		fput(sbi->dif0.file);
+	kfree(sbi->volume_name);
+	kfree(sbi);
+}
+
+static void erofs_fc_free(struct fs_context *fc)
+{
+	struct erofs_sb_info *sbi = fc->s_fs_info;
+
+	/*
+	 * Freed here only if an error occurs before the sb is set up; at that
+	 * point no block-backed device has been claimed (that happens in
+	 * fill_super), so the NULL sb never reaches fs_bdev_file_release().
+	 */
+	if (sbi)
+		erofs_sb_free(sbi, NULL);
+}
+
+static const struct fs_context_operations erofs_context_ops = {
+	.parse_param	= erofs_fc_parse_param,
+	.get_tree       = erofs_fc_get_tree,
+	.reconfigure    = erofs_fc_reconfigure,
+	.free		= erofs_fc_free,
+};
+
+static int erofs_init_fs_context(struct fs_context *fc)
+{
+	struct erofs_sb_info *sbi;
+
+	sbi = kzalloc_obj(*sbi);
+	if (!sbi)
+		return -ENOMEM;
+
+	sbi->devs = kzalloc_obj(struct erofs_dev_context);
+	if (!sbi->devs) {
+		kfree(sbi);
+		return -ENOMEM;
+	}
+	fc->s_fs_info = sbi;
+
+	idr_init(&sbi->devs->tree);
+	init_rwsem(&sbi->devs->rwsem);
+	erofs_default_options(sbi);
+	fc->ops = &erofs_context_ops;
+	return 0;
+}
+
+static void erofs_drop_internal_inodes(struct erofs_sb_info *sbi)
+{
+	iput(sbi->packed_inode);
+	sbi->packed_inode = NULL;
+	iput(sbi->metabox_inode);
+	sbi->metabox_inode = NULL;
+	iput(sbi->managed_cache);
+	sbi->managed_cache = NULL;
+}
+
+static void erofs_kill_sb(struct super_block *sb)
+{
+	struct erofs_sb_info *sbi = EROFS_SB(sb);
+
+	if (sbi->dif0.file)
+		kill_anon_super(sb);
+	else
+		kill_block_super(sb);
+	erofs_drop_internal_inodes(sbi);
+	fs_put_dax(sbi->dif0.dax_dev, NULL);
+	erofs_sb_free(sbi, sb);
+	sb->s_fs_info = NULL;
+}
+
+static void erofs_put_super(struct super_block *sb)
+{
+	struct erofs_sb_info *const sbi = EROFS_SB(sb);
+
+	erofs_unregister_sysfs(sb);
+	erofs_shrinker_unregister(sb);
+	erofs_xattr_prefixes_cleanup(sb);
+	erofs_drop_internal_inodes(sbi);
+	erofs_free_dev_context(sbi->devs, sb);
+	sbi->devs = NULL;
+}
+
+static struct file_system_type erofs_fs_type = {
+	.owner          = THIS_MODULE,
+	.name           = "erofs",
+	.init_fs_context = erofs_init_fs_context,
+	.kill_sb        = erofs_kill_sb,
+	.fs_flags       = FS_REQUIRES_DEV | FS_ALLOW_IDMAP,
+};
+MODULE_ALIAS_FS("erofs");
+
+#ifdef CONFIG_EROFS_FS_PAGE_CACHE_SHARE
+static void erofs_free_anon_inode(struct inode *inode)
+{
+	struct erofs_inode *vi = EROFS_I(inode);
+
+	kfree(vi->fingerprint.opaque);
+	kmem_cache_free(erofs_inode_cachep, vi);
+}
+
+static const struct super_operations erofs_anon_sops = {
+	.alloc_inode = erofs_alloc_inode,
+	.drop_inode = inode_just_drop,
+	.free_inode = erofs_free_anon_inode,
+};
+
+static int erofs_anon_init_fs_context(struct fs_context *fc)
+{
+	struct pseudo_fs_context *ctx;
+
+	ctx = init_pseudo(fc, EROFS_SUPER_MAGIC);
+	if (!ctx)
+		return -ENOMEM;
+	ctx->ops = &erofs_anon_sops;
+	return 0;
+}
+
+struct file_system_type erofs_anon_fs_type = {
+	.name           = "pseudo_erofs",
+	.init_fs_context = erofs_anon_init_fs_context,
+	.kill_sb        = kill_anon_super,
+};
+#endif
+
+static int __init erofs_module_init(void)
+{
+	int err;
+
+	erofs_check_ondisk_layout_definitions();
+
+	erofs_inode_cachep = kmem_cache_create("erofs_inode",
+			sizeof(struct erofs_inode), 0,
+			SLAB_RECLAIM_ACCOUNT | SLAB_ACCOUNT,
+			erofs_inode_init_once);
+	if (!erofs_inode_cachep)
+		return -ENOMEM;
+
+	err = erofs_init_shrinker();
+	if (err)
+		goto shrinker_err;
+
+	err = z_erofs_init_subsystem();
+	if (err)
+		goto zip_err;
+
+	err = erofs_init_sysfs();
+	if (err)
+		goto sysfs_err;
+
+	err = erofs_init_ishare();
+	if (err)
+		goto ishare_err;
+
+	err = register_filesystem(&erofs_fs_type);
+	if (err)
+		goto fs_err;
+
+	return 0;
+
+fs_err:
+	erofs_exit_ishare();
+ishare_err:
+	erofs_exit_sysfs();
+sysfs_err:
+	z_erofs_exit_subsystem();
+zip_err:
+	erofs_exit_shrinker();
+shrinker_err:
+	kmem_cache_destroy(erofs_inode_cachep);
+	return err;
+}
+
+static void __exit erofs_module_exit(void)
+{
+	unregister_filesystem(&erofs_fs_type);
+	erofs_exit_ishare();
+
+	/* ensure all delayed rcu free inodes & pclusters are flushed */
+	rcu_barrier();
+
+	erofs_exit_sysfs();
+	z_erofs_exit_subsystem();
+	erofs_exit_shrinker();
+	kmem_cache_destroy(erofs_inode_cachep);
+}
+
+static int erofs_statfs(struct dentry *dentry, struct kstatfs *buf)
+{
+	struct super_block *sb = dentry->d_sb;
+	struct erofs_sb_info *sbi = EROFS_SB(sb);
+
+	buf->f_type = sb->s_magic;
+	buf->f_bsize = sb->s_blocksize;
+	buf->f_blocks = sbi->total_blocks;
+	buf->f_bfree = buf->f_bavail = 0;
+	buf->f_files = ULLONG_MAX;
+	buf->f_ffree = ULLONG_MAX - sbi->inos;
+	buf->f_namelen = EROFS_NAME_LEN;
+
+	if (uuid_is_null(&sb->s_uuid))
+		buf->f_fsid = u64_to_fsid(!sb->s_bdev ? 0 :
+				huge_encode_dev(sb->s_bdev->bd_dev));
+	else
+		buf->f_fsid = uuid_to_fsid(sb->s_uuid.b);
+	return 0;
+}
+
+static int erofs_show_options(struct seq_file *seq, struct dentry *root)
+{
+	struct erofs_sb_info *sbi = EROFS_SB(root->d_sb);
+	struct erofs_mount_opts *opt = &sbi->opt;
+
+	if (IS_ENABLED(CONFIG_EROFS_FS_XATTR))
+		seq_puts(seq, test_opt(opt, XATTR_USER) ?
+				",user_xattr" : ",nouser_xattr");
+	if (IS_ENABLED(CONFIG_EROFS_FS_POSIX_ACL))
+		seq_puts(seq, test_opt(opt, POSIX_ACL) ? ",acl" : ",noacl");
+	if (IS_ENABLED(CONFIG_EROFS_FS_ZIP) && sbi->available_compr_algs)
+		seq_printf(seq, ",cache_strategy=%s",
+			  erofs_param_cache_strategy[opt->cache_strategy].name);
+	if (test_opt(opt, DAX_ALWAYS))
+		seq_puts(seq, ",dax=always");
+	if (test_opt(opt, DAX_NEVER))
+		seq_puts(seq, ",dax=never");
+	if (erofs_is_fileio_mode(sbi) && test_opt(opt, DIRECT_IO))
+		seq_puts(seq, ",directio");
+	if (sbi->dif0.fsoff)
+		seq_printf(seq, ",fsoffset=%llu", sbi->dif0.fsoff);
+	if (test_opt(opt, INODE_SHARE))
+		seq_puts(seq, ",inode_share");
+	return 0;
+}
+
+static void erofs_evict_inode(struct inode *inode)
+{
+	if (IS_DAX(inode))
+		dax_break_layout_final(inode);
+	erofs_ishare_free_inode(inode);
+	truncate_inode_pages_final(&inode->i_data);
+	clear_inode(inode);
+}
+
+const struct super_operations erofs_sops = {
+	.put_super = erofs_put_super,
+	.alloc_inode = erofs_alloc_inode,
+	.free_inode = erofs_free_inode,
+	.evict_inode = erofs_evict_inode,
+	.statfs = erofs_statfs,
+	.show_options = erofs_show_options,
+};
+
+module_init(erofs_module_init);
+module_exit(erofs_module_exit);
+
+MODULE_DESCRIPTION("Enhanced ROM File System");
+MODULE_AUTHOR("Gao Xiang, Chao Yu, Miao Xie, CONSUMER BG, HUAWEI Inc.");
+MODULE_LICENSE("GPL");
diff --git a/fs/erofs/sysfs.c b/fs/erofs/sysfs.c
new file mode 100644
index 000000000..6734483a4
--- /dev/null
+++ b/fs/erofs/sysfs.c
@@ -0,0 +1,302 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (C), 2008-2021, OPPO Mobile Comm Corp., Ltd.
+ *             https://www.oppo.com/
+ */
+#include <linux/sysfs.h>
+#include <linux/kobject.h>
+
+#include "internal.h"
+#include "compress.h"
+
+enum {
+	attr_feature,
+	attr_drop_caches,
+	attr_pointer_ui,
+	attr_pointer_bool,
+	attr_accel,
+};
+
+enum {
+	struct_erofs_sb_info,
+	struct_erofs_mount_opts,
+};
+
+struct erofs_attr {
+	struct attribute attr;
+	short attr_id;
+	int struct_type, offset;
+};
+
+#define EROFS_ATTR(_name, _mode, _id)					\
+static struct erofs_attr erofs_attr_##_name = {				\
+	.attr = {.name = __stringify(_name), .mode = _mode },		\
+	.attr_id = attr_##_id,						\
+}
+#define EROFS_ATTR_FUNC(_name, _mode)	EROFS_ATTR(_name, _mode, _name)
+#define EROFS_ATTR_FEATURE(_name)	EROFS_ATTR(_name, 0444, feature)
+
+#define EROFS_ATTR_OFFSET(_name, _mode, _id, _struct)	\
+static struct erofs_attr erofs_attr_##_name = {			\
+	.attr = {.name = __stringify(_name), .mode = _mode },	\
+	.attr_id = attr_##_id,					\
+	.struct_type = struct_##_struct,			\
+	.offset = offsetof(struct _struct, _name),\
+}
+
+#define EROFS_ATTR_RW(_name, _id, _struct)	\
+	EROFS_ATTR_OFFSET(_name, 0644, _id, _struct)
+
+#define EROFS_RO_ATTR(_name, _id, _struct)	\
+	EROFS_ATTR_OFFSET(_name, 0444, _id, _struct)
+
+#define EROFS_ATTR_RW_UI(_name, _struct)	\
+	EROFS_ATTR_RW(_name, pointer_ui, _struct)
+
+#define EROFS_ATTR_RW_BOOL(_name, _struct)	\
+	EROFS_ATTR_RW(_name, pointer_bool, _struct)
+
+#define ATTR_LIST(name) (&erofs_attr_##name.attr)
+
+#ifdef CONFIG_EROFS_FS_ZIP
+EROFS_ATTR_RW_UI(sync_decompress, erofs_sb_info);
+EROFS_ATTR_FUNC(drop_caches, 0200);
+#endif
+#ifdef CONFIG_EROFS_FS_ZIP_ACCEL
+EROFS_ATTR_FUNC(accel, 0644);
+#endif
+EROFS_ATTR_RW_UI(dir_ra_bytes, erofs_sb_info);
+
+static struct attribute *erofs_sb_attrs[] = {
+#ifdef CONFIG_EROFS_FS_ZIP
+	ATTR_LIST(sync_decompress),
+	ATTR_LIST(drop_caches),
+#endif
+	ATTR_LIST(dir_ra_bytes),
+	NULL,
+};
+ATTRIBUTE_GROUPS(erofs_sb);
+
+static struct attribute *erofs_attrs[] = {
+#ifdef CONFIG_EROFS_FS_ZIP_ACCEL
+	ATTR_LIST(accel),
+#endif
+	NULL,
+};
+ATTRIBUTE_GROUPS(erofs);
+
+/* Features this copy of erofs supports */
+EROFS_ATTR_FEATURE(compr_cfgs);
+EROFS_ATTR_FEATURE(big_pcluster);
+EROFS_ATTR_FEATURE(chunked_file);
+EROFS_ATTR_FEATURE(device_table);
+EROFS_ATTR_FEATURE(compr_head2);
+EROFS_ATTR_FEATURE(sb_chksum);
+EROFS_ATTR_FEATURE(ztailpacking);
+EROFS_ATTR_FEATURE(fragments);
+EROFS_ATTR_FEATURE(dedupe);
+EROFS_ATTR_FEATURE(48bit);
+EROFS_ATTR_FEATURE(metabox);
+
+static struct attribute *erofs_feat_attrs[] = {
+	ATTR_LIST(compr_cfgs),
+	ATTR_LIST(big_pcluster),
+	ATTR_LIST(chunked_file),
+	ATTR_LIST(device_table),
+	ATTR_LIST(compr_head2),
+	ATTR_LIST(sb_chksum),
+	ATTR_LIST(ztailpacking),
+	ATTR_LIST(fragments),
+	ATTR_LIST(dedupe),
+	ATTR_LIST(48bit),
+	ATTR_LIST(metabox),
+	NULL,
+};
+ATTRIBUTE_GROUPS(erofs_feat);
+
+static unsigned char *__struct_ptr(struct erofs_sb_info *sbi,
+					  int struct_type, int offset)
+{
+	if (struct_type == struct_erofs_sb_info)
+		return (unsigned char *)sbi + offset;
+	if (struct_type == struct_erofs_mount_opts)
+		return (unsigned char *)&sbi->opt + offset;
+	return NULL;
+}
+
+static ssize_t erofs_attr_show(struct kobject *kobj,
+				struct attribute *attr, char *buf)
+{
+	struct erofs_sb_info *sbi = container_of(kobj, struct erofs_sb_info,
+						s_kobj);
+	struct erofs_attr *a = container_of(attr, struct erofs_attr, attr);
+	unsigned char *ptr = __struct_ptr(sbi, a->struct_type, a->offset);
+
+	switch (a->attr_id) {
+	case attr_feature:
+		return sysfs_emit(buf, "supported\n");
+	case attr_pointer_ui:
+		if (!ptr)
+			return 0;
+		return sysfs_emit(buf, "%u\n", *(unsigned int *)ptr);
+	case attr_pointer_bool:
+		if (!ptr)
+			return 0;
+		return sysfs_emit(buf, "%d\n", *(bool *)ptr);
+	case attr_accel:
+		return z_erofs_crypto_show_engines(buf, PAGE_SIZE, '\n');
+	}
+	return 0;
+}
+
+static ssize_t erofs_attr_store(struct kobject *kobj, struct attribute *attr,
+				const char *buf, size_t len)
+{
+	struct erofs_sb_info *sbi = container_of(kobj, struct erofs_sb_info,
+						s_kobj);
+	struct erofs_attr *a = container_of(attr, struct erofs_attr, attr);
+	unsigned char *ptr = __struct_ptr(sbi, a->struct_type, a->offset);
+	unsigned long t;
+	int ret;
+
+	switch (a->attr_id) {
+	case attr_pointer_ui:
+		if (!ptr)
+			return 0;
+		ret = kstrtoul(skip_spaces(buf), 0, &t);
+		if (ret)
+			return ret;
+		if (t != (unsigned int)t)
+			return -ERANGE;
+		if (IS_ENABLED(CONFIG_EROFS_FS_ZIP) &&
+		    !strcmp(a->attr.name, "sync_decompress") &&
+		    (t > EROFS_SYNC_DECOMPRESS_FORCE_OFF))
+			return -EINVAL;
+		*(unsigned int *)ptr = t;
+		return len;
+	case attr_pointer_bool:
+		if (!ptr)
+			return 0;
+		ret = kstrtoul(skip_spaces(buf), 0, &t);
+		if (ret)
+			return ret;
+		if (t != 0 && t != 1)
+			return -EINVAL;
+		*(bool *)ptr = !!t;
+		return len;
+#ifdef CONFIG_EROFS_FS_ZIP
+	case attr_drop_caches:
+		ret = kstrtoul(skip_spaces(buf), 0, &t);
+		if (ret)
+			return ret;
+		if (t < 1 || t > 3)
+			return -EINVAL;
+
+		if (t & 2)
+			z_erofs_shrink_scan(sbi, ~0UL);
+		if (t & 1)
+			invalidate_mapping_pages(MNGD_MAPPING(sbi), 0, -1);
+		return len;
+#endif
+#ifdef CONFIG_EROFS_FS_ZIP_ACCEL
+	case attr_accel:
+		buf = skip_spaces(buf);
+		z_erofs_crypto_disable_all_engines();
+		while (*buf) {
+			t = strcspn(buf, "\n");
+			ret = z_erofs_crypto_enable_engine(buf, t);
+			if (ret < 0)
+				return ret;
+			buf += buf[t] != '\0' ? t + 1 : t;
+		}
+		return len;
+#endif
+	}
+	return 0;
+}
+
+static void erofs_sb_release(struct kobject *kobj)
+{
+	struct erofs_sb_info *sbi = container_of(kobj, struct erofs_sb_info,
+						 s_kobj);
+	complete(&sbi->s_kobj_unregister);
+}
+
+static const struct sysfs_ops erofs_attr_ops = {
+	.show	= erofs_attr_show,
+	.store	= erofs_attr_store,
+};
+
+static const struct kobj_type erofs_sb_ktype = {
+	.default_groups = erofs_sb_groups,
+	.sysfs_ops	= &erofs_attr_ops,
+	.release	= erofs_sb_release,
+};
+
+static const struct kobj_type erofs_ktype = {
+	.default_groups = erofs_groups,
+	.sysfs_ops	= &erofs_attr_ops,
+};
+
+static struct kset erofs_root = {
+	.kobj	= {.ktype = &erofs_ktype},
+};
+
+static const struct kobj_type erofs_feat_ktype = {
+	.default_groups = erofs_feat_groups,
+	.sysfs_ops	= &erofs_attr_ops,
+};
+
+static struct kobject erofs_feat = {
+	.kset	= &erofs_root,
+};
+
+int erofs_register_sysfs(struct super_block *sb)
+{
+	struct erofs_sb_info *sbi = EROFS_SB(sb);
+	int err;
+
+	sbi->s_kobj.kset = &erofs_root;
+	init_completion(&sbi->s_kobj_unregister);
+	err = kobject_init_and_add(&sbi->s_kobj, &erofs_sb_ktype, NULL, "%s",
+				   sb->s_sysfs_name);
+	if (err) {
+		kobject_put(&sbi->s_kobj);
+		wait_for_completion(&sbi->s_kobj_unregister);
+	}
+	return err;
+}
+
+void erofs_unregister_sysfs(struct super_block *sb)
+{
+	struct erofs_sb_info *sbi = EROFS_SB(sb);
+
+	if (sbi->s_kobj.state_in_sysfs) {
+		kobject_del(&sbi->s_kobj);
+		kobject_put(&sbi->s_kobj);
+		wait_for_completion(&sbi->s_kobj_unregister);
+	}
+}
+
+void erofs_exit_sysfs(void)
+{
+	kobject_put(&erofs_feat);
+	kset_unregister(&erofs_root);
+}
+
+int __init erofs_init_sysfs(void)
+{
+	int ret;
+
+	kobject_set_name(&erofs_root.kobj, "erofs");
+	erofs_root.kobj.parent = fs_kobj;
+	ret = kset_register(&erofs_root);
+	if (!ret) {
+		ret = kobject_init_and_add(&erofs_feat, &erofs_feat_ktype,
+					   NULL, "features");
+		if (!ret)
+			return 0;
+		erofs_exit_sysfs();
+	}
+	return ret;
+}
diff --git a/fs/erofs/xattr.c b/fs/erofs/xattr.c
new file mode 100644
index 000000000..df7ea0195
--- /dev/null
+++ b/fs/erofs/xattr.c
@@ -0,0 +1,649 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2017-2018 HUAWEI, Inc.
+ *             https://www.huawei.com/
+ * Copyright (C) 2021-2022, Alibaba Cloud
+ */
+#include <linux/security.h>
+#include <linux/xxhash.h>
+#include "xattr.h"
+
+struct erofs_xattr_iter {
+	struct super_block *sb;
+	struct erofs_buf buf;
+	erofs_off_t pos;
+	void *kaddr;
+
+	char *buffer;
+	int buffer_size, buffer_ofs;
+
+	/* getxattr */
+	int index, infix_len;
+	struct qstr name;
+
+	/* listxattr */
+	struct dentry *dentry;
+};
+
+static const char *erofs_xattr_prefix(unsigned int idx, struct dentry *dentry);
+
+static int erofs_init_inode_xattrs(struct inode *inode)
+{
+	struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
+	struct erofs_inode *vi = EROFS_I(inode);
+	struct super_block *sb = inode->i_sb;
+	const struct erofs_xattr_ibody_header *ih;
+	__le32 *xattr_id;
+	erofs_off_t pos;
+	unsigned int i;
+	int ret = 0;
+
+	if (!vi->xattr_isize)
+		return -ENODATA;
+
+	/* the most case is that xattrs of this inode are initialized. */
+	if (test_bit(EROFS_I_EA_INITED_BIT, &vi->flags)) {
+		/*
+		 * paired with smp_mb() at the end of the function to ensure
+		 * fields will only be observed after the bit is set.
+		 */
+		smp_mb();
+		return 0;
+	}
+	if (wait_on_bit_lock(&vi->flags, EROFS_I_BL_XATTR_BIT, TASK_KILLABLE))
+		return -ERESTARTSYS;
+
+	/* someone has initialized xattrs for us? */
+	if (test_bit(EROFS_I_EA_INITED_BIT, &vi->flags))
+		goto out_unlock;
+
+	/*
+	 * bypass all xattr operations if ->xattr_isize is not greater than
+	 * sizeof(struct erofs_xattr_ibody_header), in detail:
+	 * 1) it is not enough to contain erofs_xattr_ibody_header then
+	 *    ->xattr_isize should be 0 (it means no xattr);
+	 * 2) it is just to contain erofs_xattr_ibody_header, which is on-disk
+	 *    undefined right now (maybe use later with some new sb feature).
+	 */
+	if (vi->xattr_isize == sizeof(struct erofs_xattr_ibody_header)) {
+		erofs_err(sb, "xattr_isize %d of nid %llu is not supported yet",
+			  vi->xattr_isize, vi->nid);
+		ret = -EOPNOTSUPP;
+		goto out_unlock;
+	} else if (vi->xattr_isize < sizeof(struct erofs_xattr_ibody_header)) {
+		erofs_err(sb, "bogus xattr ibody @ nid %llu", vi->nid);
+		DBG_BUGON(1);
+		ret = -EFSCORRUPTED;
+		goto out_unlock;
+	}
+
+	pos = erofs_iloc(inode) + vi->inode_isize;
+	ih = erofs_read_metabuf(&buf, sb, pos, erofs_inode_in_metabox(inode));
+	if (IS_ERR(ih)) {
+		ret = PTR_ERR(ih);
+		goto out_unlock;
+	}
+	vi->xattr_name_filter = le32_to_cpu(ih->h_name_filter);
+	vi->xattr_shared_count = ih->h_shared_count;
+	if ((u32)vi->xattr_shared_count * sizeof(__le32) >
+	    vi->xattr_isize - sizeof(struct erofs_xattr_ibody_header)) {
+		erofs_err(sb, "invalid h_shared_count %u @ nid %llu",
+			  vi->xattr_shared_count, vi->nid);
+		ret = -EFSCORRUPTED;
+		goto out_unlock;
+	}
+	vi->xattr_shared_xattrs = kmalloc_objs(uint, vi->xattr_shared_count);
+	if (!vi->xattr_shared_xattrs) {
+		ret = -ENOMEM;
+		goto out_unlock;
+	}
+
+	/* skip the ibody header and read the shared xattr array */
+	pos += sizeof(struct erofs_xattr_ibody_header);
+	for (i = 0; i < vi->xattr_shared_count; ++i) {
+		xattr_id = erofs_bread(&buf, pos + i * sizeof(__le32), true);
+		if (IS_ERR(xattr_id)) {
+			kfree(vi->xattr_shared_xattrs);
+			vi->xattr_shared_xattrs = NULL;
+			ret = PTR_ERR(xattr_id);
+			goto out_unlock;
+		}
+		vi->xattr_shared_xattrs[i] = le32_to_cpu(*xattr_id);
+	}
+
+	/* paired with smp_mb() at the beginning of the function. */
+	smp_mb();
+	set_bit(EROFS_I_EA_INITED_BIT, &vi->flags);
+out_unlock:
+	erofs_put_metabuf(&buf);
+	clear_and_wake_up_bit(EROFS_I_BL_XATTR_BIT, &vi->flags);
+	return ret;
+}
+
+static int erofs_xattr_copy_to_buffer(struct erofs_xattr_iter *it,
+				      unsigned int len)
+{
+	unsigned int slice, processed;
+	struct super_block *sb = it->sb;
+
+	for (processed = 0; processed < len; processed += slice) {
+		it->kaddr = erofs_bread(&it->buf, it->pos, true);
+		if (IS_ERR(it->kaddr))
+			return PTR_ERR(it->kaddr);
+
+		slice = min_t(unsigned int, sb->s_blocksize -
+				erofs_blkoff(sb, it->pos), len - processed);
+		memcpy(it->buffer + it->buffer_ofs, it->kaddr, slice);
+		it->buffer_ofs += slice;
+		it->pos += slice;
+	}
+	return 0;
+}
+
+static int erofs_listxattr_foreach(struct erofs_xattr_iter *it)
+{
+	struct erofs_xattr_entry entry;
+	unsigned int base_index, name_total, prefix_len, infix_len = 0;
+	const char *prefix, *infix = NULL;
+	int err;
+
+	/* 1. handle xattr entry */
+	entry = *(struct erofs_xattr_entry *)it->kaddr;
+	it->pos += sizeof(struct erofs_xattr_entry);
+
+	base_index = entry.e_name_index;
+	if (entry.e_name_index & EROFS_XATTR_LONG_PREFIX) {
+		struct erofs_sb_info *sbi = EROFS_SB(it->sb);
+		struct erofs_xattr_prefix_item *pf = sbi->xattr_prefixes +
+			(entry.e_name_index & EROFS_XATTR_LONG_PREFIX_MASK);
+
+		if (pf >= sbi->xattr_prefixes + sbi->xattr_prefix_count)
+			return 0;
+		infix = pf->prefix->infix;
+		infix_len = pf->infix_len;
+		base_index = pf->prefix->base_index;
+	}
+
+	prefix = erofs_xattr_prefix(base_index, it->dentry);
+	if (!prefix)
+		return 0;
+	prefix_len = strlen(prefix);
+	name_total = prefix_len + infix_len + entry.e_name_len + 1;
+
+	if (!it->buffer) {
+		it->buffer_ofs += name_total;
+		return 0;
+	}
+
+	if (it->buffer_ofs + name_total > it->buffer_size)
+		return -ERANGE;
+
+	memcpy(it->buffer + it->buffer_ofs, prefix, prefix_len);
+	memcpy(it->buffer + it->buffer_ofs + prefix_len, infix, infix_len);
+	it->buffer_ofs += prefix_len + infix_len;
+
+	/* 2. handle xattr name */
+	err = erofs_xattr_copy_to_buffer(it, entry.e_name_len);
+	if (err)
+		return err;
+
+	it->buffer[it->buffer_ofs++] = '\0';
+	return 0;
+}
+
+static int erofs_getxattr_foreach(struct erofs_xattr_iter *it)
+{
+	struct super_block *sb = it->sb;
+	struct erofs_xattr_entry entry;
+	unsigned int slice, processed, value_sz;
+
+	/* 1. handle xattr entry */
+	entry = *(struct erofs_xattr_entry *)it->kaddr;
+	it->pos += sizeof(struct erofs_xattr_entry);
+	value_sz = le16_to_cpu(entry.e_value_size);
+
+	/* should also match the infix for long name prefixes */
+	if (entry.e_name_index & EROFS_XATTR_LONG_PREFIX) {
+		struct erofs_sb_info *sbi = EROFS_SB(sb);
+		struct erofs_xattr_prefix_item *pf = sbi->xattr_prefixes +
+			(entry.e_name_index & EROFS_XATTR_LONG_PREFIX_MASK);
+
+		if (pf >= sbi->xattr_prefixes + sbi->xattr_prefix_count)
+			return -ENODATA;
+
+		if (it->index != pf->prefix->base_index ||
+		    it->name.len != entry.e_name_len + pf->infix_len)
+			return -ENODATA;
+
+		if (memcmp(it->name.name, pf->prefix->infix, pf->infix_len))
+			return -ENODATA;
+
+		it->infix_len = pf->infix_len;
+	} else {
+		if (it->index != entry.e_name_index ||
+		    it->name.len != entry.e_name_len)
+			return -ENODATA;
+
+		it->infix_len = 0;
+	}
+
+	/* 2. handle xattr name */
+	for (processed = 0; processed < entry.e_name_len; processed += slice) {
+		it->kaddr = erofs_bread(&it->buf, it->pos, true);
+		if (IS_ERR(it->kaddr))
+			return PTR_ERR(it->kaddr);
+
+		slice = min_t(unsigned int,
+				sb->s_blocksize - erofs_blkoff(sb, it->pos),
+				entry.e_name_len - processed);
+		if (memcmp(it->name.name + it->infix_len + processed,
+			   it->kaddr, slice))
+			return -ENODATA;
+		it->pos += slice;
+	}
+
+	/* 3. handle xattr value */
+	if (!it->buffer) {
+		it->buffer_ofs = value_sz;
+		return 0;
+	}
+
+	if (it->buffer_size < value_sz)
+		return -ERANGE;
+
+	return erofs_xattr_copy_to_buffer(it, value_sz);
+}
+
+static int erofs_xattr_iter_inline(struct erofs_xattr_iter *it,
+				   struct inode *inode, bool getxattr)
+{
+	struct erofs_inode *const vi = EROFS_I(inode);
+	unsigned int xattr_header_sz, remaining, entry_sz;
+	erofs_off_t next_pos;
+	int ret;
+
+	xattr_header_sz = sizeof(struct erofs_xattr_ibody_header) +
+			  sizeof(u32) * vi->xattr_shared_count;
+	if (xattr_header_sz >= vi->xattr_isize) {
+		DBG_BUGON(xattr_header_sz > vi->xattr_isize);
+		return -ENODATA;
+	}
+
+	ret = erofs_init_metabuf(&it->buf, it->sb, erofs_inode_in_metabox(inode));
+	if (ret)
+		return ret;
+	remaining = vi->xattr_isize - xattr_header_sz;
+	it->pos = erofs_iloc(inode) + vi->inode_isize + xattr_header_sz;
+
+	while (remaining) {
+		it->kaddr = erofs_bread(&it->buf, it->pos, true);
+		if (IS_ERR(it->kaddr))
+			return PTR_ERR(it->kaddr);
+
+		entry_sz = erofs_xattr_entry_size(it->kaddr);
+		/* xattr on-disk corruption: xattr entry beyond xattr_isize */
+		if (remaining < entry_sz) {
+			DBG_BUGON(1);
+			return -EFSCORRUPTED;
+		}
+		remaining -= entry_sz;
+		next_pos = it->pos + entry_sz;
+
+		if (getxattr)
+			ret = erofs_getxattr_foreach(it);
+		else
+			ret = erofs_listxattr_foreach(it);
+		if ((getxattr && ret != -ENODATA) || (!getxattr && ret))
+			break;
+
+		it->pos = next_pos;
+	}
+	return ret;
+}
+
+static int erofs_xattr_iter_shared(struct erofs_xattr_iter *it,
+				   struct inode *inode, bool getxattr)
+{
+	struct erofs_inode *const vi = EROFS_I(inode);
+	struct super_block *const sb = it->sb;
+	struct erofs_sb_info *sbi = EROFS_SB(sb);
+	unsigned int i = 0;
+	int ret;
+
+	ret = erofs_init_metabuf(&it->buf, sb,
+				 erofs_sb_has_shared_ea_in_metabox(sbi));
+	if (ret)
+		return ret;
+
+	while (i < vi->xattr_shared_count) {
+		it->pos = erofs_pos(sb, sbi->xattr_blkaddr) +
+				vi->xattr_shared_xattrs[i++] * sizeof(__le32);
+		it->kaddr = erofs_bread(&it->buf, it->pos, true);
+		if (IS_ERR(it->kaddr))
+			return PTR_ERR(it->kaddr);
+
+		if (getxattr)
+			ret = erofs_getxattr_foreach(it);
+		else
+			ret = erofs_listxattr_foreach(it);
+		if ((getxattr && ret != -ENODATA) || (!getxattr && ret))
+			break;
+	}
+	return i ? ret : -ENODATA;
+}
+
+static int erofs_getxattr(struct inode *inode, int index, const char *name,
+			  void *buffer, size_t buffer_size)
+{
+	int ret;
+	unsigned int hashbit;
+	struct erofs_xattr_iter it;
+	struct erofs_inode *vi = EROFS_I(inode);
+	struct erofs_sb_info *sbi = EROFS_SB(inode->i_sb);
+
+	if (!name)
+		return -EINVAL;
+
+	ret = erofs_init_inode_xattrs(inode);
+	if (ret)
+		return ret;
+
+	/* reserved flag is non-zero if there's any change of on-disk format */
+	if (erofs_sb_has_xattr_filter(sbi) && !sbi->xattr_filter_reserved) {
+		hashbit = xxh32(name, strlen(name),
+				EROFS_XATTR_FILTER_SEED + index);
+		hashbit &= EROFS_XATTR_FILTER_BITS - 1;
+		if (vi->xattr_name_filter & (1U << hashbit))
+			return -ENODATA;
+	}
+
+	it.index = index;
+	it.name = QSTR(name);
+	if (it.name.len > EROFS_NAME_LEN)
+		return -ERANGE;
+
+	it.sb = inode->i_sb;
+	it.buf = __EROFS_BUF_INITIALIZER;
+	it.buffer = buffer;
+	it.buffer_size = buffer_size;
+	it.buffer_ofs = 0;
+
+	ret = erofs_xattr_iter_inline(&it, inode, true);
+	if (ret == -ENODATA)
+		ret = erofs_xattr_iter_shared(&it, inode, true);
+	erofs_put_metabuf(&it.buf);
+	return ret ? ret : it.buffer_ofs;
+}
+
+ssize_t erofs_listxattr(struct dentry *dentry, char *buffer, size_t buffer_size)
+{
+	int ret;
+	struct erofs_xattr_iter it;
+	struct inode *inode = d_inode(dentry);
+
+	ret = erofs_init_inode_xattrs(inode);
+	if (ret == -ENODATA)
+		return 0;
+	if (ret)
+		return ret;
+
+	it.sb = dentry->d_sb;
+	it.buf = __EROFS_BUF_INITIALIZER;
+	it.dentry = dentry;
+	it.buffer = buffer;
+	it.buffer_size = buffer_size;
+	it.buffer_ofs = 0;
+
+	ret = erofs_xattr_iter_inline(&it, inode, false);
+	if (!ret || ret == -ENODATA)
+		ret = erofs_xattr_iter_shared(&it, inode, false);
+	if (ret == -ENODATA)
+		ret = 0;
+	erofs_put_metabuf(&it.buf);
+	return ret ? ret : it.buffer_ofs;
+}
+
+static bool erofs_xattr_user_list(struct dentry *dentry)
+{
+	return test_opt(&EROFS_SB(dentry->d_sb)->opt, XATTR_USER);
+}
+
+static bool erofs_xattr_trusted_list(struct dentry *dentry)
+{
+	return capable(CAP_SYS_ADMIN);
+}
+
+static int erofs_xattr_generic_get(const struct xattr_handler *handler,
+				   struct dentry *unused, struct inode *inode,
+				   const char *name, void *buffer, size_t size)
+{
+	if (handler->flags == EROFS_XATTR_INDEX_USER &&
+	    !test_opt(&EROFS_I_SB(inode)->opt, XATTR_USER))
+		return -EOPNOTSUPP;
+
+	return erofs_getxattr(inode, handler->flags, name, buffer, size);
+}
+
+static const struct xattr_handler erofs_xattr_user_handler = {
+	.prefix	= XATTR_USER_PREFIX,
+	.flags	= EROFS_XATTR_INDEX_USER,
+	.list	= erofs_xattr_user_list,
+	.get	= erofs_xattr_generic_get,
+};
+
+static const struct xattr_handler erofs_xattr_trusted_handler = {
+	.prefix	= XATTR_TRUSTED_PREFIX,
+	.flags	= EROFS_XATTR_INDEX_TRUSTED,
+	.list	= erofs_xattr_trusted_list,
+	.get	= erofs_xattr_generic_get,
+};
+
+#ifdef CONFIG_EROFS_FS_SECURITY
+static const struct xattr_handler erofs_xattr_security_handler = {
+	.prefix	= XATTR_SECURITY_PREFIX,
+	.flags	= EROFS_XATTR_INDEX_SECURITY,
+	.get	= erofs_xattr_generic_get,
+};
+#endif
+
+const struct xattr_handler * const erofs_xattr_handlers[] = {
+	&erofs_xattr_user_handler,
+	&erofs_xattr_trusted_handler,
+#ifdef CONFIG_EROFS_FS_SECURITY
+	&erofs_xattr_security_handler,
+#endif
+	NULL,
+};
+
+static const char *erofs_xattr_prefix(unsigned int idx, struct dentry *dentry)
+{
+	static const struct xattr_handler * const xattr_handler_map[] = {
+		[EROFS_XATTR_INDEX_USER] = &erofs_xattr_user_handler,
+#ifdef CONFIG_EROFS_FS_POSIX_ACL
+		[EROFS_XATTR_INDEX_POSIX_ACL_ACCESS] = &nop_posix_acl_access,
+		[EROFS_XATTR_INDEX_POSIX_ACL_DEFAULT] = &nop_posix_acl_default,
+#endif
+		[EROFS_XATTR_INDEX_TRUSTED] = &erofs_xattr_trusted_handler,
+#ifdef CONFIG_EROFS_FS_SECURITY
+		[EROFS_XATTR_INDEX_SECURITY] = &erofs_xattr_security_handler,
+#endif
+	};
+	const struct xattr_handler *handler = NULL;
+
+	if (idx && idx < ARRAY_SIZE(xattr_handler_map)) {
+		handler = xattr_handler_map[idx];
+		if (xattr_handler_can_list(handler, dentry))
+			return xattr_prefix(handler);
+	}
+	return NULL;
+}
+
+void erofs_xattr_prefixes_cleanup(struct super_block *sb)
+{
+	struct erofs_sb_info *sbi = EROFS_SB(sb);
+	int i;
+
+	if (sbi->xattr_prefixes) {
+		for (i = 0; i < sbi->xattr_prefix_count; i++)
+			kfree(sbi->xattr_prefixes[i].prefix);
+		kfree(sbi->xattr_prefixes);
+		sbi->xattr_prefixes = NULL;
+	}
+}
+
+int erofs_xattr_prefixes_init(struct super_block *sb)
+{
+	struct erofs_sb_info *sbi = EROFS_SB(sb);
+	struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
+	erofs_off_t pos = (erofs_off_t)sbi->xattr_prefix_start << 2;
+	struct erofs_xattr_prefix_item *pfs;
+	int ret = 0, i, len;
+	bool plain = erofs_sb_has_plain_xattr_pfx(sbi);
+
+	if (!sbi->xattr_prefix_count)
+		return 0;
+
+	pfs = kzalloc_objs(*pfs, sbi->xattr_prefix_count);
+	if (!pfs)
+		return -ENOMEM;
+
+	if (!plain) {
+		if (erofs_sb_has_metabox(sbi))
+			(void)erofs_init_metabuf(&buf, sb, true);
+		else if (sbi->packed_inode)
+			buf.mapping = sbi->packed_inode->i_mapping;
+		else
+			plain = true;
+	}
+	if (plain)
+		(void)erofs_init_metabuf(&buf, sb, false);
+
+	for (i = 0; i < sbi->xattr_prefix_count; i++) {
+		void *ptr = erofs_read_metadata(sb, &buf, &pos, &len);
+
+		if (IS_ERR(ptr)) {
+			ret = PTR_ERR(ptr);
+			break;
+		} else if (len < sizeof(*pfs->prefix) ||
+			   len > EROFS_NAME_LEN + sizeof(*pfs->prefix)) {
+			kfree(ptr);
+			ret = -EFSCORRUPTED;
+			break;
+		}
+		pfs[i].prefix = ptr;
+		pfs[i].infix_len = len - sizeof(struct erofs_xattr_long_prefix);
+	}
+
+	erofs_put_metabuf(&buf);
+	if (!ret && erofs_sb_has_ishare_xattrs(sbi)) {
+		struct erofs_xattr_prefix_item *pf = pfs + sbi->ishare_xattr_prefix_id;
+		struct erofs_xattr_long_prefix *newpfx;
+
+		newpfx = krealloc(pf->prefix,
+			sizeof(*newpfx) + pf->infix_len + 1, GFP_KERNEL);
+		if (newpfx) {
+			newpfx->infix[pf->infix_len] = '\0';
+			pf->prefix = newpfx;
+		} else {
+			ret = -ENOMEM;
+		}
+	}
+	sbi->xattr_prefixes = pfs;
+	if (ret)
+		erofs_xattr_prefixes_cleanup(sb);
+	return ret;
+}
+
+#ifdef CONFIG_EROFS_FS_POSIX_ACL
+struct posix_acl *erofs_get_acl(struct inode *inode, int type, bool rcu)
+{
+	struct posix_acl *acl;
+	int prefix, rc;
+	char *value = NULL;
+
+	if (rcu)
+		return ERR_PTR(-ECHILD);
+
+	switch (type) {
+	case ACL_TYPE_ACCESS:
+		prefix = EROFS_XATTR_INDEX_POSIX_ACL_ACCESS;
+		break;
+	case ACL_TYPE_DEFAULT:
+		prefix = EROFS_XATTR_INDEX_POSIX_ACL_DEFAULT;
+		break;
+	default:
+		return ERR_PTR(-EINVAL);
+	}
+
+	rc = erofs_getxattr(inode, prefix, "", NULL, 0);
+	if (rc > 0) {
+		value = kmalloc(rc, GFP_KERNEL);
+		if (!value)
+			return ERR_PTR(-ENOMEM);
+		rc = erofs_getxattr(inode, prefix, "", value, rc);
+	}
+
+	if (rc == -ENODATA)
+		acl = NULL;
+	else if (rc < 0)
+		acl = ERR_PTR(rc);
+	else
+		acl = posix_acl_from_xattr(&init_user_ns, value, rc);
+	kfree(value);
+	return acl;
+}
+
+bool erofs_inode_has_noacl(struct inode *inode, void *kaddr, unsigned int ofs)
+{
+	static const unsigned int bitmask =
+		BIT(21) |	/* system.posix_acl_default */
+		BIT(30);	/* system.posix_acl_access */
+	struct erofs_sb_info *sbi = EROFS_I_SB(inode);
+	const struct erofs_xattr_ibody_header *ih = kaddr + ofs;
+
+	if (EROFS_I(inode)->xattr_isize < sizeof(*ih))
+		return true;
+
+	if (erofs_sb_has_xattr_filter(sbi) && !sbi->xattr_filter_reserved &&
+	    !check_add_overflow(ofs, sizeof(*ih), &ofs) &&
+	    ofs <= i_blocksize(inode)) {
+		if ((le32_to_cpu(ih->h_name_filter) & bitmask) == bitmask)
+			return true;
+	}
+	return false;
+}
+#endif
+
+#ifdef CONFIG_EROFS_FS_PAGE_CACHE_SHARE
+int erofs_xattr_fill_inode_fingerprint(struct erofs_inode_fingerprint *fp,
+				       struct inode *inode, const char *domain_id)
+{
+	struct erofs_sb_info *sbi = EROFS_SB(inode->i_sb);
+	struct erofs_xattr_prefix_item *prefix;
+	const char *infix;
+	int valuelen, base_index;
+
+	if (!test_opt(&sbi->opt, INODE_SHARE))
+		return -EOPNOTSUPP;
+	if (!sbi->xattr_prefixes)
+		return -EINVAL;
+	prefix = sbi->xattr_prefixes + sbi->ishare_xattr_prefix_id;
+	infix = prefix->prefix->infix;
+	base_index = prefix->prefix->base_index;
+	valuelen = erofs_getxattr(inode, base_index, infix, NULL, 0);
+	if (valuelen <= 0 || valuelen > (1 << sbi->blkszbits))
+		return -EFSCORRUPTED;
+	fp->size = valuelen + (domain_id ? strlen(domain_id) : 0);
+	fp->opaque = kmalloc(fp->size, GFP_KERNEL);
+	if (!fp->opaque)
+		return -ENOMEM;
+	if (valuelen != erofs_getxattr(inode, base_index, infix,
+				       fp->opaque, valuelen)) {
+		kfree(fp->opaque);
+		fp->opaque = NULL;
+		return -EFSCORRUPTED;
+	}
+	memcpy(fp->opaque + valuelen, domain_id, fp->size - valuelen);
+	return 0;
+}
+#endif
diff --git a/fs/erofs/xattr.h b/fs/erofs/xattr.h
new file mode 100644
index 000000000..4465f7018
--- /dev/null
+++ b/fs/erofs/xattr.h
@@ -0,0 +1,36 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (C) 2017-2018 HUAWEI, Inc.
+ *             https://www.huawei.com/
+ */
+#ifndef __EROFS_XATTR_H
+#define __EROFS_XATTR_H
+
+#include "internal.h"
+#include <linux/posix_acl_xattr.h>
+#include <linux/xattr.h>
+
+#ifdef CONFIG_EROFS_FS_XATTR
+extern const struct xattr_handler * const erofs_xattr_handlers[];
+
+int erofs_xattr_prefixes_init(struct super_block *sb);
+void erofs_xattr_prefixes_cleanup(struct super_block *sb);
+ssize_t erofs_listxattr(struct dentry *, char *, size_t);
+#else
+static inline int erofs_xattr_prefixes_init(struct super_block *sb) { return 0; }
+static inline void erofs_xattr_prefixes_cleanup(struct super_block *sb) {}
+
+#define erofs_listxattr (NULL)
+#define erofs_xattr_handlers (NULL)
+#endif	/* !CONFIG_EROFS_FS_XATTR */
+
+#ifdef CONFIG_EROFS_FS_POSIX_ACL
+struct posix_acl *erofs_get_acl(struct inode *inode, int type, bool rcu);
+#else
+#define erofs_get_acl	(NULL)
+#endif
+
+int erofs_xattr_fill_inode_fingerprint(struct erofs_inode_fingerprint *fp,
+				       struct inode *inode, const char *domain_id);
+bool erofs_inode_has_noacl(struct inode *inode, void *kaddr, unsigned int ofs);
+#endif
diff --git a/fs/erofs/zdata.c b/fs/erofs/zdata.c
new file mode 100644
index 000000000..e1e25ca0d
--- /dev/null
+++ b/fs/erofs/zdata.c
@@ -0,0 +1,1935 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2018 HUAWEI, Inc.
+ *             https://www.huawei.com/
+ * Copyright (C) 2022 Alibaba Cloud
+ */
+#include "compress.h"
+#include <linux/psi.h>
+#include <linux/cpuhotplug.h>
+#include <trace/events/erofs.h>
+
+#define Z_EROFS_MAX_SYNC_DECOMPRESS_BYTES	12288
+#define Z_EROFS_PCLUSTER_MAX_PAGES	(Z_EROFS_PCLUSTER_MAX_SIZE / PAGE_SIZE)
+#define Z_EROFS_INLINE_BVECS		2
+
+struct z_erofs_bvec {
+	struct page *page;
+	int offset;
+	unsigned int end;
+};
+
+#define __Z_EROFS_BVSET(name, total) \
+struct name { \
+	/* point to the next page which contains the following bvecs */ \
+	struct page *nextpage; \
+	struct z_erofs_bvec bvec[total]; \
+}
+__Z_EROFS_BVSET(z_erofs_bvset,);
+__Z_EROFS_BVSET(z_erofs_bvset_inline, Z_EROFS_INLINE_BVECS);
+
+/*
+ * Structure fields follow one of the following exclusion rules.
+ *
+ * I: Modifiable by initialization/destruction paths and read-only
+ *    for everyone else;
+ *
+ * L: Field should be protected by the pcluster lock;
+ *
+ * A: Field should be accessed / updated in atomic for parallelized code.
+ */
+struct z_erofs_pcluster {
+	struct mutex lock;
+	struct lockref lockref;
+
+	/* A: point to next chained pcluster or TAILs */
+	struct z_erofs_pcluster *next;
+
+	/* I: start physical position of this pcluster */
+	erofs_off_t pos;
+
+	/* L: the maximum decompression size of this round */
+	unsigned int length;
+
+	/* L: total number of bvecs */
+	unsigned int vcnt;
+
+	/* I: pcluster size (compressed size) in bytes */
+	unsigned int pclustersize;
+
+	/* I: page offset of start position of decompression */
+	unsigned short pageofs_out;
+
+	/* I: page offset of inline compressed data */
+	unsigned short pageofs_in;
+
+	union {
+		/* L: inline a certain number of bvec for bootstrap */
+		struct z_erofs_bvset_inline bvset;
+
+		/* I: can be used to free the pcluster by RCU. */
+		struct rcu_head rcu;
+	};
+
+	/* I: compression algorithm format */
+	unsigned char algorithmformat;
+
+	/* I: whether compressed data is in-lined or not */
+	bool from_meta;
+
+	/* L: whether partial decompression or not */
+	bool partial;
+
+	/* L: whether extra buffer allocations are best-effort */
+	bool besteffort;
+
+	/* A: compressed bvecs (can be cached or inplaced pages) */
+	struct z_erofs_bvec compressed_bvecs[];
+};
+
+/* the end of a chain of pclusters */
+#define Z_EROFS_PCLUSTER_TAIL           ((void *) 0x700 + POISON_POINTER_DELTA)
+
+struct z_erofs_decompressqueue {
+	struct super_block *sb;
+	struct z_erofs_pcluster *head;
+	atomic_t pending_bios;
+
+	union {
+		struct completion done;
+		struct work_struct work;
+		struct kthread_work kthread_work;
+	} u;
+	bool eio, sync;
+};
+
+static inline unsigned int z_erofs_pclusterpages(struct z_erofs_pcluster *pcl)
+{
+	return PAGE_ALIGN(pcl->pageofs_in + pcl->pclustersize) >> PAGE_SHIFT;
+}
+
+static bool erofs_folio_is_managed(struct erofs_sb_info *sbi, struct folio *fo)
+{
+	return fo->mapping == MNGD_MAPPING(sbi);
+}
+
+#define Z_EROFS_ONSTACK_PAGES		32
+
+/*
+ * since pclustersize is variable for big pcluster feature, introduce slab
+ * pools implementation for different pcluster sizes.
+ */
+struct z_erofs_pcluster_slab {
+	struct kmem_cache *slab;
+	unsigned int maxpages;
+	char name[48];
+};
+
+#define _PCLP(n) { .maxpages = n }
+
+static struct z_erofs_pcluster_slab pcluster_pool[] __read_mostly = {
+	_PCLP(1),
+	_PCLP(4),
+#if Z_EROFS_PCLUSTER_MAX_PAGES > 16
+	_PCLP(16),
+#endif
+#if Z_EROFS_PCLUSTER_MAX_PAGES > 64
+	_PCLP(64),
+#endif
+#if Z_EROFS_PCLUSTER_MAX_PAGES > 128
+	_PCLP(128),
+#endif
+	_PCLP(Z_EROFS_PCLUSTER_MAX_PAGES + 1)
+};
+
+struct z_erofs_bvec_iter {
+	struct page *bvpage;
+	struct z_erofs_bvset *bvset;
+	unsigned int nr, cur;
+};
+
+static struct page *z_erofs_bvec_iter_end(struct z_erofs_bvec_iter *iter)
+{
+	if (iter->bvpage)
+		kunmap_local(iter->bvset);
+	return iter->bvpage;
+}
+
+static struct page *z_erofs_bvset_flip(struct z_erofs_bvec_iter *iter)
+{
+	unsigned long base = (unsigned long)((struct z_erofs_bvset *)0)->bvec;
+	/* have to access nextpage in advance, otherwise it will be unmapped */
+	struct page *nextpage = iter->bvset->nextpage;
+	struct page *oldpage;
+
+	DBG_BUGON(!nextpage);
+	oldpage = z_erofs_bvec_iter_end(iter);
+	iter->bvpage = nextpage;
+	iter->bvset = kmap_local_page(nextpage);
+	iter->nr = (PAGE_SIZE - base) / sizeof(struct z_erofs_bvec);
+	iter->cur = 0;
+	return oldpage;
+}
+
+static void z_erofs_bvec_iter_begin(struct z_erofs_bvec_iter *iter,
+				    struct z_erofs_bvset_inline *bvset,
+				    unsigned int bootstrap_nr,
+				    unsigned int cur)
+{
+	*iter = (struct z_erofs_bvec_iter) {
+		.nr = bootstrap_nr,
+		.bvset = (struct z_erofs_bvset *)bvset,
+	};
+
+	while (cur > iter->nr) {
+		cur -= iter->nr;
+		z_erofs_bvset_flip(iter);
+	}
+	iter->cur = cur;
+}
+
+static int z_erofs_bvec_enqueue(struct z_erofs_bvec_iter *iter,
+				struct z_erofs_bvec *bvec,
+				struct page **candidate_bvpage,
+				struct page **pagepool)
+{
+	if (iter->cur >= iter->nr) {
+		struct page *nextpage = *candidate_bvpage;
+
+		if (!nextpage) {
+			nextpage = __erofs_allocpage(pagepool, GFP_KERNEL,
+					true);
+			if (!nextpage)
+				return -ENOMEM;
+			set_page_private(nextpage, Z_EROFS_SHORTLIVED_PAGE);
+		}
+		DBG_BUGON(iter->bvset->nextpage);
+		iter->bvset->nextpage = nextpage;
+		z_erofs_bvset_flip(iter);
+
+		iter->bvset->nextpage = NULL;
+		*candidate_bvpage = NULL;
+	}
+	iter->bvset->bvec[iter->cur++] = *bvec;
+	return 0;
+}
+
+static void z_erofs_bvec_dequeue(struct z_erofs_bvec_iter *iter,
+				 struct z_erofs_bvec *bvec,
+				 struct page **old_bvpage)
+{
+	if (iter->cur == iter->nr)
+		*old_bvpage = z_erofs_bvset_flip(iter);
+	else
+		*old_bvpage = NULL;
+	*bvec = iter->bvset->bvec[iter->cur++];
+}
+
+static void z_erofs_destroy_pcluster_pool(void)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(pcluster_pool); ++i) {
+		if (!pcluster_pool[i].slab)
+			continue;
+		kmem_cache_destroy(pcluster_pool[i].slab);
+		pcluster_pool[i].slab = NULL;
+	}
+}
+
+static int z_erofs_create_pcluster_pool(void)
+{
+	struct z_erofs_pcluster_slab *pcs;
+	struct z_erofs_pcluster *a;
+	unsigned int size;
+
+	for (pcs = pcluster_pool;
+	     pcs < pcluster_pool + ARRAY_SIZE(pcluster_pool); ++pcs) {
+		size = struct_size(a, compressed_bvecs, pcs->maxpages);
+
+		sprintf(pcs->name, "erofs_pcluster-%u", pcs->maxpages);
+		pcs->slab = kmem_cache_create(pcs->name, size, 0,
+					      SLAB_RECLAIM_ACCOUNT, NULL);
+		if (pcs->slab)
+			continue;
+
+		z_erofs_destroy_pcluster_pool();
+		return -ENOMEM;
+	}
+	return 0;
+}
+
+static struct z_erofs_pcluster *z_erofs_alloc_pcluster(unsigned int size)
+{
+	unsigned int nrpages = PAGE_ALIGN(size) >> PAGE_SHIFT;
+	struct z_erofs_pcluster_slab *pcs = pcluster_pool;
+
+	for (; pcs < pcluster_pool + ARRAY_SIZE(pcluster_pool); ++pcs) {
+		struct z_erofs_pcluster *pcl;
+
+		if (nrpages > pcs->maxpages)
+			continue;
+
+		pcl = kmem_cache_zalloc(pcs->slab, GFP_KERNEL);
+		if (!pcl)
+			return ERR_PTR(-ENOMEM);
+		return pcl;
+	}
+	return ERR_PTR(-EINVAL);
+}
+
+static void z_erofs_free_pcluster(struct z_erofs_pcluster *pcl)
+{
+	unsigned int pclusterpages = z_erofs_pclusterpages(pcl);
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(pcluster_pool); ++i) {
+		struct z_erofs_pcluster_slab *pcs = pcluster_pool + i;
+
+		if (pclusterpages > pcs->maxpages)
+			continue;
+
+		kmem_cache_free(pcs->slab, pcl);
+		return;
+	}
+	DBG_BUGON(1);
+}
+
+static struct workqueue_struct *z_erofs_workqueue __read_mostly;
+
+#ifdef CONFIG_EROFS_FS_PCPU_KTHREAD
+static struct kthread_worker __rcu **z_erofs_pcpu_workers;
+static atomic_t erofs_percpu_workers_initialized = ATOMIC_INIT(0);
+
+static void erofs_destroy_percpu_workers(void)
+{
+	struct kthread_worker *worker;
+	unsigned int cpu;
+
+	for_each_possible_cpu(cpu) {
+		worker = rcu_dereference_protected(
+					z_erofs_pcpu_workers[cpu], 1);
+		rcu_assign_pointer(z_erofs_pcpu_workers[cpu], NULL);
+		if (worker)
+			kthread_destroy_worker(worker);
+	}
+	kfree(z_erofs_pcpu_workers);
+}
+
+static struct kthread_worker *erofs_init_percpu_worker(int cpu)
+{
+	struct kthread_worker *worker =
+		kthread_run_worker_on_cpu(cpu, 0, "erofs_worker/%u");
+
+	if (IS_ERR(worker))
+		return worker;
+	if (IS_ENABLED(CONFIG_EROFS_FS_PCPU_KTHREAD_HIPRI))
+		sched_set_fifo_low(worker->task);
+	return worker;
+}
+
+static int erofs_init_percpu_workers(void)
+{
+	struct kthread_worker *worker;
+	unsigned int cpu;
+
+	z_erofs_pcpu_workers = kzalloc_objs(struct kthread_worker *,
+					    num_possible_cpus(), GFP_ATOMIC);
+	if (!z_erofs_pcpu_workers)
+		return -ENOMEM;
+
+	for_each_online_cpu(cpu) {	/* could miss cpu{off,on}line? */
+		worker = erofs_init_percpu_worker(cpu);
+		if (!IS_ERR(worker))
+			rcu_assign_pointer(z_erofs_pcpu_workers[cpu], worker);
+	}
+	return 0;
+}
+
+#ifdef CONFIG_HOTPLUG_CPU
+static DEFINE_SPINLOCK(z_erofs_pcpu_worker_lock);
+static enum cpuhp_state erofs_cpuhp_state;
+
+static int erofs_cpu_online(unsigned int cpu)
+{
+	struct kthread_worker *worker, *old;
+
+	worker = erofs_init_percpu_worker(cpu);
+	if (IS_ERR(worker))
+		return PTR_ERR(worker);
+
+	spin_lock(&z_erofs_pcpu_worker_lock);
+	old = rcu_dereference_protected(z_erofs_pcpu_workers[cpu],
+			lockdep_is_held(&z_erofs_pcpu_worker_lock));
+	if (!old)
+		rcu_assign_pointer(z_erofs_pcpu_workers[cpu], worker);
+	spin_unlock(&z_erofs_pcpu_worker_lock);
+	if (old)
+		kthread_destroy_worker(worker);
+	return 0;
+}
+
+static int erofs_cpu_offline(unsigned int cpu)
+{
+	struct kthread_worker *worker;
+
+	spin_lock(&z_erofs_pcpu_worker_lock);
+	worker = rcu_dereference_protected(z_erofs_pcpu_workers[cpu],
+			lockdep_is_held(&z_erofs_pcpu_worker_lock));
+	rcu_assign_pointer(z_erofs_pcpu_workers[cpu], NULL);
+	spin_unlock(&z_erofs_pcpu_worker_lock);
+
+	synchronize_rcu();
+	if (worker)
+		kthread_destroy_worker(worker);
+	return 0;
+}
+
+static int erofs_cpu_hotplug_init(void)
+{
+	int state;
+
+	state = cpuhp_setup_state_nocalls(CPUHP_AP_ONLINE_DYN,
+			"fs/erofs:online", erofs_cpu_online, erofs_cpu_offline);
+	if (state < 0)
+		return state;
+
+	erofs_cpuhp_state = state;
+	return 0;
+}
+
+static void erofs_cpu_hotplug_destroy(void)
+{
+	if (erofs_cpuhp_state)
+		cpuhp_remove_state_nocalls(erofs_cpuhp_state);
+}
+#else /* !CONFIG_HOTPLUG_CPU  */
+static inline int erofs_cpu_hotplug_init(void) { return 0; }
+static inline void erofs_cpu_hotplug_destroy(void) {}
+#endif/* CONFIG_HOTPLUG_CPU */
+static int z_erofs_init_pcpu_workers(struct super_block *sb)
+{
+	int err;
+
+	if (atomic_xchg(&erofs_percpu_workers_initialized, 1))
+		return 0;
+
+	err = erofs_init_percpu_workers();
+	if (err) {
+		erofs_err(sb, "per-cpu workers: failed to allocate.");
+		goto err_init_percpu_workers;
+	}
+
+	err = erofs_cpu_hotplug_init();
+	if (err < 0) {
+		erofs_err(sb, "per-cpu workers: failed CPU hotplug init.");
+		goto err_cpuhp_init;
+	}
+	erofs_info(sb, "initialized per-cpu workers successfully.");
+	return err;
+
+err_cpuhp_init:
+	erofs_destroy_percpu_workers();
+err_init_percpu_workers:
+	atomic_set(&erofs_percpu_workers_initialized, 0);
+	return err;
+}
+
+static void z_erofs_destroy_pcpu_workers(void)
+{
+	if (!atomic_xchg(&erofs_percpu_workers_initialized, 0))
+		return;
+	erofs_cpu_hotplug_destroy();
+	erofs_destroy_percpu_workers();
+}
+#else /* !CONFIG_EROFS_FS_PCPU_KTHREAD */
+static inline int z_erofs_init_pcpu_workers(struct super_block *sb) { return 0; }
+static inline void z_erofs_destroy_pcpu_workers(void) {}
+#endif/* CONFIG_EROFS_FS_PCPU_KTHREAD */
+
+void z_erofs_exit_subsystem(void)
+{
+	z_erofs_destroy_pcpu_workers();
+	destroy_workqueue(z_erofs_workqueue);
+	z_erofs_destroy_pcluster_pool();
+	z_erofs_crypto_disable_all_engines();
+	z_erofs_exit_decompressor();
+}
+
+int __init z_erofs_init_subsystem(void)
+{
+	int err = z_erofs_init_decompressor();
+
+	if (err)
+		goto err_decompressor;
+
+	err = z_erofs_create_pcluster_pool();
+	if (err)
+		goto err_pcluster_pool;
+
+	z_erofs_workqueue = alloc_workqueue("erofs_worker",
+			WQ_UNBOUND | WQ_HIGHPRI, num_possible_cpus());
+	if (!z_erofs_workqueue) {
+		err = -ENOMEM;
+		goto err_workqueue_init;
+	}
+
+	return err;
+
+err_workqueue_init:
+	z_erofs_destroy_pcluster_pool();
+err_pcluster_pool:
+	z_erofs_exit_decompressor();
+err_decompressor:
+	return err;
+}
+
+enum z_erofs_pclustermode {
+	/* It has previously been linked into another processing chain */
+	Z_EROFS_PCLUSTER_INFLIGHT,
+	/*
+	 * A weaker form of Z_EROFS_PCLUSTER_FOLLOWED; the difference is that it
+	 * may be dispatched to the bypass queue later due to uptodated managed
+	 * folios.  All file-backed folios related to this pcluster cannot be
+	 * reused for in-place I/O (or bvpage) since the pcluster may be decoded
+	 * in a separate queue (and thus out of order).
+	 */
+	Z_EROFS_PCLUSTER_FOLLOWED_NOINPLACE,
+	/*
+	 * The pcluster has just been linked to our processing chain.
+	 * File-backed folios (except for the head page) related to it can be
+	 * used for in-place I/O (or bvpage).
+	 */
+	Z_EROFS_PCLUSTER_FOLLOWED,
+};
+
+struct z_erofs_frontend {
+	struct inode *inode, *sharedinode;
+	struct erofs_map_blocks map;
+	struct z_erofs_bvec_iter biter;
+
+	struct page *pagepool;
+	struct page *candidate_bvpage;
+	struct z_erofs_pcluster *pcl, *head;
+	enum z_erofs_pclustermode mode;
+
+	erofs_off_t headoffset;
+
+	/* a pointer used to pick up inplace I/O pages */
+	unsigned int icur;
+};
+
+#define Z_EROFS_DEFINE_FRONTEND(fe, i, si, ho) struct z_erofs_frontend fe = { \
+	.inode = i, .sharedinode = si, .head = Z_EROFS_PCLUSTER_TAIL, \
+	.mode = Z_EROFS_PCLUSTER_FOLLOWED, .headoffset = ho }
+
+static bool z_erofs_should_alloc_cache(struct z_erofs_frontend *fe)
+{
+	unsigned int cachestrategy = EROFS_I_SB(fe->inode)->opt.cache_strategy;
+
+	if (cachestrategy <= EROFS_ZIP_CACHE_DISABLED)
+		return false;
+
+	if (fe->map.m_flags & EROFS_MAP_PARTIAL_MAPPED)
+		return true;
+
+	if (cachestrategy >= EROFS_ZIP_CACHE_READAROUND &&
+	    fe->map.m_la < fe->headoffset)
+		return true;
+
+	return false;
+}
+
+static void z_erofs_bind_cache(struct z_erofs_frontend *fe)
+{
+	struct address_space *mc = MNGD_MAPPING(EROFS_I_SB(fe->inode));
+	struct z_erofs_pcluster *pcl = fe->pcl;
+	unsigned int pclusterpages = z_erofs_pclusterpages(pcl);
+	bool shouldalloc = z_erofs_should_alloc_cache(fe);
+	pgoff_t poff = pcl->pos >> PAGE_SHIFT;
+	bool may_bypass = true;
+	/* Optimistic allocation, as in-place I/O can be used as a fallback */
+	gfp_t gfp = (mapping_gfp_mask(mc) & ~__GFP_DIRECT_RECLAIM) |
+			__GFP_NOMEMALLOC | __GFP_NORETRY | __GFP_NOWARN;
+	struct folio *folio, *newfolio;
+	unsigned int i;
+
+	if (i_blocksize(fe->inode) != PAGE_SIZE ||
+	    fe->mode < Z_EROFS_PCLUSTER_FOLLOWED)
+		return;
+
+	for (i = 0; i < pclusterpages; ++i) {
+		/* Inaccurate check w/o locking to avoid unneeded lookups */
+		if (READ_ONCE(pcl->compressed_bvecs[i].page))
+			continue;
+
+		folio = filemap_get_folio(mc, poff + i);
+		if (IS_ERR(folio)) {
+			may_bypass = false;
+			if (!shouldalloc)
+				continue;
+
+			/*
+			 * Allocate a managed folio for cached I/O, or it may be
+			 * then filled with a file-backed folio for in-place I/O
+			 */
+			newfolio = filemap_alloc_folio(gfp, 0, NULL);
+			if (!newfolio)
+				continue;
+			newfolio->private = Z_EROFS_PREALLOCATED_FOLIO;
+			folio = NULL;
+		}
+		spin_lock(&pcl->lockref.lock);
+		if (!pcl->compressed_bvecs[i].page) {
+			pcl->compressed_bvecs[i].page =
+				folio_page(folio ?: newfolio, 0);
+			spin_unlock(&pcl->lockref.lock);
+			continue;
+		}
+		spin_unlock(&pcl->lockref.lock);
+		folio_put(folio ?: newfolio);
+	}
+
+	/*
+	 * Don't perform in-place I/O if all compressed pages are available in
+	 * the managed cache, as the pcluster can be moved to the bypass queue.
+	 */
+	if (may_bypass)
+		fe->mode = Z_EROFS_PCLUSTER_FOLLOWED_NOINPLACE;
+}
+
+/* (erofs_shrinker) disconnect cached encoded data with pclusters */
+static int erofs_try_to_free_all_cached_folios(struct erofs_sb_info *sbi,
+					       struct z_erofs_pcluster *pcl)
+{
+	unsigned int pclusterpages = z_erofs_pclusterpages(pcl);
+	struct folio *folio;
+	int i;
+
+	DBG_BUGON(pcl->from_meta);
+	/* Each cached folio contains one page unless bs > ps is supported */
+	for (i = 0; i < pclusterpages; ++i) {
+		if (pcl->compressed_bvecs[i].page) {
+			folio = page_folio(pcl->compressed_bvecs[i].page);
+			/* Avoid reclaiming or migrating this folio */
+			if (!folio_trylock(folio))
+				return -EBUSY;
+
+			DBG_BUGON(!erofs_folio_is_managed(sbi, folio));
+			pcl->compressed_bvecs[i].page = NULL;
+			folio_detach_private(folio);
+			folio_unlock(folio);
+		}
+	}
+	return 0;
+}
+
+static bool z_erofs_cache_release_folio(struct folio *folio, gfp_t gfp)
+{
+	struct z_erofs_pcluster *pcl = folio_get_private(folio);
+	struct z_erofs_bvec *bvec = pcl->compressed_bvecs;
+	struct z_erofs_bvec *end = bvec + z_erofs_pclusterpages(pcl);
+	bool ret;
+
+	if (!folio_test_private(folio))
+		return true;
+
+	ret = false;
+	spin_lock(&pcl->lockref.lock);
+	if (pcl->lockref.count <= 0) {
+		DBG_BUGON(pcl->from_meta);
+		for (; bvec < end; ++bvec) {
+			if (bvec->page && page_folio(bvec->page) == folio) {
+				bvec->page = NULL;
+				folio_detach_private(folio);
+				ret = true;
+				break;
+			}
+		}
+	}
+	spin_unlock(&pcl->lockref.lock);
+	return ret;
+}
+
+/*
+ * It will be called only on inode eviction. In case that there are still some
+ * decompression requests in progress, wait with rescheduling for a bit here.
+ * An extra lock could be introduced instead but it seems unnecessary.
+ */
+static void z_erofs_cache_invalidate_folio(struct folio *folio,
+					   size_t offset, size_t length)
+{
+	const size_t stop = length + offset;
+
+	/* Check for potential overflow in debug mode */
+	DBG_BUGON(stop > folio_size(folio) || stop < length);
+
+	if (offset == 0 && stop == folio_size(folio))
+		while (!z_erofs_cache_release_folio(folio, 0))
+			cond_resched();
+}
+
+static const struct address_space_operations z_erofs_cache_aops = {
+	.release_folio = z_erofs_cache_release_folio,
+	.invalidate_folio = z_erofs_cache_invalidate_folio,
+};
+
+int z_erofs_init_super(struct super_block *sb)
+{
+	int err;
+
+	err = z_erofs_init_pcpu_workers(sb);
+	if (err)
+		return err;
+
+	err = erofs_setup_managed_cache(sb);
+	if (err)
+		return err;
+	EROFS_SB(sb)->managed_cache->i_mapping->a_ops = &z_erofs_cache_aops;
+	xa_init(&EROFS_SB(sb)->managed_pslots);
+	return 0;
+}
+
+/* callers must be with pcluster lock held */
+static int z_erofs_attach_page(struct z_erofs_frontend *fe,
+			       struct z_erofs_bvec *bvec, bool exclusive)
+{
+	struct z_erofs_pcluster *pcl = fe->pcl;
+	int ret;
+
+	if (exclusive) {
+		/* Inplace I/O is limited to one page for uncompressed data */
+		if (pcl->algorithmformat < Z_EROFS_COMPRESSION_MAX ||
+		    fe->icur <= 1) {
+			/* Try to prioritize inplace I/O here */
+			spin_lock(&pcl->lockref.lock);
+			while (fe->icur > 0) {
+				if (pcl->compressed_bvecs[--fe->icur].page)
+					continue;
+				pcl->compressed_bvecs[fe->icur] = *bvec;
+				spin_unlock(&pcl->lockref.lock);
+				return 0;
+			}
+			spin_unlock(&pcl->lockref.lock);
+		}
+
+		/* otherwise, check if it can be used as a bvpage */
+		if (fe->mode >= Z_EROFS_PCLUSTER_FOLLOWED &&
+		    !fe->candidate_bvpage)
+			fe->candidate_bvpage = bvec->page;
+	}
+	ret = z_erofs_bvec_enqueue(&fe->biter, bvec, &fe->candidate_bvpage,
+				   &fe->pagepool);
+	fe->pcl->vcnt += (ret >= 0);
+	return ret;
+}
+
+static bool z_erofs_get_pcluster(struct z_erofs_pcluster *pcl)
+{
+	if (lockref_get_not_zero(&pcl->lockref))
+		return true;
+
+	spin_lock(&pcl->lockref.lock);
+	if (lockref_is_dead(&pcl->lockref)) {
+		spin_unlock(&pcl->lockref.lock);
+		return false;
+	}
+
+	if (!pcl->lockref.count++)
+		atomic_long_dec(&erofs_global_shrink_cnt);
+	spin_unlock(&pcl->lockref.lock);
+	return true;
+}
+
+static int z_erofs_register_pcluster(struct z_erofs_frontend *fe)
+{
+	struct erofs_map_blocks *map = &fe->map;
+	struct super_block *sb = fe->inode->i_sb;
+	struct erofs_sb_info *sbi = EROFS_SB(sb);
+	struct z_erofs_pcluster *pcl, *pre;
+	unsigned int pageofs_in;
+	int err;
+
+	pageofs_in = erofs_blkoff(sb, map->m_pa);
+	pcl = z_erofs_alloc_pcluster(pageofs_in + map->m_plen);
+	if (IS_ERR(pcl))
+		return PTR_ERR(pcl);
+
+	lockref_init(&pcl->lockref); /* one ref for this request */
+	pcl->algorithmformat = map->m_algorithmformat;
+	pcl->pclustersize = map->m_plen;
+	pcl->length = 0;
+	pcl->partial = true;
+	pcl->next = fe->head;
+	pcl->pos = map->m_pa;
+	pcl->pageofs_in = pageofs_in;
+	pcl->pageofs_out = map->m_la & ~PAGE_MASK;
+	pcl->from_meta = map->m_flags & EROFS_MAP_META;
+	fe->mode = Z_EROFS_PCLUSTER_FOLLOWED;
+
+	/*
+	 * lock all primary followed works before visible to others
+	 * and mutex_trylock *never* fails for a new pcluster.
+	 */
+	mutex_init(&pcl->lock);
+	DBG_BUGON(!mutex_trylock(&pcl->lock));
+
+	if (!pcl->from_meta) {
+		while (1) {
+			xa_lock(&sbi->managed_pslots);
+			pre = __xa_cmpxchg(&sbi->managed_pslots, pcl->pos,
+					   NULL, pcl, GFP_KERNEL);
+			if (!pre || xa_is_err(pre) || z_erofs_get_pcluster(pre)) {
+				xa_unlock(&sbi->managed_pslots);
+				break;
+			}
+			/* try to legitimize the current in-tree one */
+			xa_unlock(&sbi->managed_pslots);
+			cond_resched();
+		}
+		if (xa_is_err(pre)) {
+			err = xa_err(pre);
+			goto err_out;
+		} else if (pre) {
+			fe->pcl = pre;
+			err = -EEXIST;
+			goto err_out;
+		}
+	}
+	fe->head = fe->pcl = pcl;
+	return 0;
+
+err_out:
+	mutex_unlock(&pcl->lock);
+	z_erofs_free_pcluster(pcl);
+	return err;
+}
+
+static int z_erofs_pcluster_begin(struct z_erofs_frontend *fe)
+{
+	struct erofs_map_blocks *map = &fe->map;
+	struct super_block *sb = fe->inode->i_sb;
+	struct z_erofs_pcluster *pcl = NULL;
+	void *ptr = NULL;
+	bool needretry;
+	int ret;
+
+	DBG_BUGON(fe->pcl);
+	/* must be Z_EROFS_PCLUSTER_TAIL or pointed to previous pcluster */
+	DBG_BUGON(!fe->head);
+
+	if (map->m_flags & EROFS_MAP_META) {
+		ret = erofs_init_metabuf(&map->buf, sb,
+					 erofs_inode_in_metabox(fe->inode));
+		if (ret)
+			return ret;
+		ptr = erofs_bread(&map->buf, map->m_pa, false);
+		if (IS_ERR(ptr)) {
+			erofs_err(sb, "failed to read inline data %pe @ pa %llu of nid %llu",
+				  ptr, map->m_pa, EROFS_I(fe->inode)->nid);
+			return PTR_ERR(ptr);
+		}
+		ptr = map->buf.page;
+	} else {
+		do {
+			rcu_read_lock();
+			pcl = xa_load(&EROFS_SB(sb)->managed_pslots, map->m_pa);
+			needretry = pcl && !z_erofs_get_pcluster(pcl);
+			rcu_read_unlock();
+		} while (needretry);
+	}
+
+	if (pcl) {
+		DBG_BUGON(map->m_pa != pcl->pos);
+		fe->pcl = pcl;
+		ret = -EEXIST;
+	} else {
+		ret = z_erofs_register_pcluster(fe);
+	}
+
+	if (ret == -EEXIST) {
+		mutex_lock(&fe->pcl->lock);
+		/* check if this pcluster hasn't been linked into any chain. */
+		if (!cmpxchg(&fe->pcl->next, NULL, fe->head)) {
+			/* .. so it can be attached to our submission chain */
+			fe->head = fe->pcl;
+			fe->mode = Z_EROFS_PCLUSTER_FOLLOWED;
+		} else {	/* otherwise, it belongs to an inflight chain */
+			fe->mode = Z_EROFS_PCLUSTER_INFLIGHT;
+		}
+	} else if (ret) {
+		return ret;
+	}
+
+	z_erofs_bvec_iter_begin(&fe->biter, &fe->pcl->bvset,
+				Z_EROFS_INLINE_BVECS, fe->pcl->vcnt);
+	if (!fe->pcl->from_meta) {
+		/* bind cache first when cached decompression is preferred */
+		z_erofs_bind_cache(fe);
+	} else {
+		folio_get(page_folio((struct page *)ptr));
+		WRITE_ONCE(fe->pcl->compressed_bvecs[0].page, ptr);
+		fe->pcl->pageofs_in = map->m_pa & ~PAGE_MASK;
+		fe->mode = Z_EROFS_PCLUSTER_FOLLOWED_NOINPLACE;
+	}
+	/* file-backed inplace I/O pages are traversed in reverse order */
+	fe->icur = z_erofs_pclusterpages(fe->pcl);
+	return 0;
+}
+
+static void z_erofs_rcu_callback(struct rcu_head *head)
+{
+	z_erofs_free_pcluster(container_of(head, struct z_erofs_pcluster, rcu));
+}
+
+static bool __erofs_try_to_release_pcluster(struct erofs_sb_info *sbi,
+					  struct z_erofs_pcluster *pcl)
+{
+	if (pcl->lockref.count)
+		return false;
+
+	/*
+	 * Note that all cached folios should be detached before deleted from
+	 * the XArray.  Otherwise some folios could be still attached to the
+	 * orphan old pcluster when the new one is available in the tree.
+	 */
+	if (erofs_try_to_free_all_cached_folios(sbi, pcl))
+		return false;
+
+	/*
+	 * It's impossible to fail after the pcluster is freezed, but in order
+	 * to avoid some race conditions, add a DBG_BUGON to observe this.
+	 */
+	DBG_BUGON(__xa_erase(&sbi->managed_pslots, pcl->pos) != pcl);
+
+	lockref_mark_dead(&pcl->lockref);
+	return true;
+}
+
+static bool erofs_try_to_release_pcluster(struct erofs_sb_info *sbi,
+					  struct z_erofs_pcluster *pcl)
+{
+	bool free;
+
+	spin_lock(&pcl->lockref.lock);
+	free = __erofs_try_to_release_pcluster(sbi, pcl);
+	spin_unlock(&pcl->lockref.lock);
+	if (free) {
+		atomic_long_dec(&erofs_global_shrink_cnt);
+		call_rcu(&pcl->rcu, z_erofs_rcu_callback);
+	}
+	return free;
+}
+
+unsigned long z_erofs_shrink_scan(struct erofs_sb_info *sbi, unsigned long nr)
+{
+	struct z_erofs_pcluster *pcl;
+	unsigned long index, freed = 0;
+
+	xa_lock(&sbi->managed_pslots);
+	xa_for_each(&sbi->managed_pslots, index, pcl) {
+		/* try to shrink each valid pcluster */
+		if (!erofs_try_to_release_pcluster(sbi, pcl))
+			continue;
+		xa_unlock(&sbi->managed_pslots);
+
+		++freed;
+		if (!--nr)
+			return freed;
+		xa_lock(&sbi->managed_pslots);
+	}
+	xa_unlock(&sbi->managed_pslots);
+	return freed;
+}
+
+static void z_erofs_put_pcluster(struct erofs_sb_info *sbi,
+		struct z_erofs_pcluster *pcl, bool try_free)
+{
+	bool free = false;
+
+	if (lockref_put_or_lock(&pcl->lockref))
+		return;
+
+	DBG_BUGON(lockref_is_dead(&pcl->lockref));
+	if (!--pcl->lockref.count) {
+		if (try_free && xa_trylock(&sbi->managed_pslots)) {
+			free = __erofs_try_to_release_pcluster(sbi, pcl);
+			xa_unlock(&sbi->managed_pslots);
+		}
+		atomic_long_add(!free, &erofs_global_shrink_cnt);
+	}
+	spin_unlock(&pcl->lockref.lock);
+	if (free)
+		call_rcu(&pcl->rcu, z_erofs_rcu_callback);
+}
+
+static void z_erofs_pcluster_end(struct z_erofs_frontend *fe)
+{
+	struct z_erofs_pcluster *pcl = fe->pcl;
+
+	if (!pcl)
+		return;
+
+	z_erofs_bvec_iter_end(&fe->biter);
+	mutex_unlock(&pcl->lock);
+
+	if (fe->candidate_bvpage)
+		fe->candidate_bvpage = NULL;
+
+	/* Drop refcount if it doesn't belong to our processing chain */
+	if (fe->mode < Z_EROFS_PCLUSTER_FOLLOWED_NOINPLACE)
+		z_erofs_put_pcluster(EROFS_I_SB(fe->inode), pcl, false);
+	fe->pcl = NULL;
+}
+
+static int z_erofs_read_fragment(struct super_block *sb, struct folio *folio,
+			unsigned int cur, unsigned int end, erofs_off_t pos)
+{
+	struct inode *packed_inode = EROFS_SB(sb)->packed_inode;
+	struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
+	unsigned int cnt;
+	u8 *src;
+
+	if (!packed_inode)
+		return -EFSCORRUPTED;
+
+	buf.mapping = packed_inode->i_mapping;
+	for (; cur < end; cur += cnt, pos += cnt) {
+		cnt = min(end - cur, sb->s_blocksize - erofs_blkoff(sb, pos));
+		src = erofs_bread(&buf, pos, true);
+		if (IS_ERR(src)) {
+			erofs_put_metabuf(&buf);
+			return PTR_ERR(src);
+		}
+		memcpy_to_folio(folio, cur, src, cnt);
+	}
+	erofs_put_metabuf(&buf);
+	return 0;
+}
+
+static int z_erofs_scan_folio(struct z_erofs_frontend *f,
+			      struct folio *folio, bool ra)
+{
+	struct inode *const inode = f->inode;
+	struct erofs_map_blocks *const map = &f->map;
+	const loff_t offset = folio_pos(folio);
+	const unsigned int bs = i_blocksize(inode);
+	unsigned int end = folio_size(folio), split = 0, cur, pgs;
+	bool tight, excl;
+	int err = 0;
+
+	tight = (bs == PAGE_SIZE);
+	erofs_onlinefolio_init(folio);
+	do {
+		if (offset + end - 1 < map->m_la ||
+		    offset + end - 1 >= map->m_la + map->m_llen) {
+			z_erofs_pcluster_end(f);
+			map->m_la = offset + end - 1;
+			map->m_llen = 0;
+			err = z_erofs_map_blocks_iter(inode, map, 0);
+			if (err)
+				break;
+		}
+
+		cur = offset > map->m_la ? 0 : map->m_la - offset;
+		pgs = round_down(cur, PAGE_SIZE);
+		/* bump split parts first to avoid several separate cases */
+		++split;
+
+		if (map->m_flags & EROFS_MAP_FRAGMENT) {
+			erofs_off_t fpos = offset + cur - map->m_la;
+
+			err = z_erofs_read_fragment(inode->i_sb, folio, cur,
+					cur + min(map->m_llen - fpos, end - cur),
+					EROFS_I(inode)->z_fragmentoff + fpos);
+			if (err)
+				break;
+			tight = false;
+		} else if (!(map->m_flags & EROFS_MAP_MAPPED)) {
+			folio_zero_segment(folio, cur, end);
+			tight = false;
+		} else {
+			if (!f->pcl) {
+				err = z_erofs_pcluster_begin(f);
+				if (err)
+					break;
+				f->pcl->besteffort |= !ra;
+			}
+
+			pgs = round_down(end - 1, PAGE_SIZE);
+			/*
+			 * Ensure this partial page belongs to this submit chain
+			 * rather than other concurrent submit chains or
+			 * noio(bypass) chains since those chains are handled
+			 * asynchronously thus it cannot be used for inplace I/O
+			 * or bvpage (should be processed in the strict order.)
+			 */
+			tight &= (f->mode >= Z_EROFS_PCLUSTER_FOLLOWED);
+			excl = false;
+			if (cur <= pgs) {
+				excl = (split <= 1) || tight;
+				cur = pgs;
+			}
+
+			err = z_erofs_attach_page(f, &((struct z_erofs_bvec) {
+				.page = folio_page(folio, pgs >> PAGE_SHIFT),
+				.offset = offset + pgs - map->m_la,
+				.end = end - pgs, }), excl);
+			if (err)
+				break;
+
+			erofs_onlinefolio_split(folio);
+			if (f->pcl->length < offset + end - map->m_la) {
+				f->pcl->length = offset + end - map->m_la;
+				f->pcl->pageofs_out = map->m_la & ~PAGE_MASK;
+			}
+			if (EROFS_MAP_FULL(map->m_flags) &&
+			    f->pcl->length == map->m_llen)
+				f->pcl->partial = false;
+		}
+		/* shorten the remaining extent to update progress */
+		map->m_llen = offset + cur - map->m_la;
+		map->m_flags |= EROFS_MAP_PARTIAL_MAPPED;
+		if (cur <= pgs) {
+			split = cur < pgs;
+			tight = (bs == PAGE_SIZE);
+		}
+	} while ((end = cur) > 0);
+	erofs_onlinefolio_end(folio, err, false);
+	return err;
+}
+
+static bool z_erofs_page_is_invalidated(struct page *page)
+{
+	return !page_folio(page)->mapping && !z_erofs_is_shortlived_page(page);
+}
+
+struct z_erofs_backend {
+	struct page *onstack_pages[Z_EROFS_ONSTACK_PAGES];
+	struct super_block *sb;
+	struct z_erofs_pcluster *pcl;
+	/* pages with the longest decompressed length for deduplication */
+	struct page **decompressed_pages;
+	/* pages to keep the compressed data */
+	struct page **compressed_pages;
+
+	struct list_head decompressed_secondary_bvecs;
+	struct page **pagepool;
+	unsigned int onstack_used, nr_pages;
+	/* indicate if temporary copies should be preserved for later use */
+	bool keepxcpy;
+};
+
+struct z_erofs_bvec_item {
+	struct z_erofs_bvec bvec;
+	struct list_head list;
+};
+
+static void z_erofs_do_decompressed_bvec(struct z_erofs_backend *be,
+					 struct z_erofs_bvec *bvec)
+{
+	int poff = bvec->offset + be->pcl->pageofs_out;
+	struct z_erofs_bvec_item *item;
+	struct page **page;
+
+	if (!(poff & ~PAGE_MASK) && (bvec->end == PAGE_SIZE ||
+			bvec->offset + bvec->end == be->pcl->length)) {
+		DBG_BUGON((poff >> PAGE_SHIFT) >= be->nr_pages);
+		page = be->decompressed_pages + (poff >> PAGE_SHIFT);
+		if (!*page) {
+			*page = bvec->page;
+			return;
+		}
+	} else {
+		be->keepxcpy = true;
+	}
+
+	/* (cold path) one pcluster is requested multiple times */
+	item = kmalloc_obj(*item, GFP_KERNEL | __GFP_NOFAIL);
+	item->bvec = *bvec;
+	list_add(&item->list, &be->decompressed_secondary_bvecs);
+}
+
+static void z_erofs_fill_other_copies(struct z_erofs_backend *be, int err)
+{
+	unsigned int off0 = be->pcl->pageofs_out;
+	struct list_head *p, *n;
+
+	list_for_each_safe(p, n, &be->decompressed_secondary_bvecs) {
+		struct z_erofs_bvec_item *bvi;
+		unsigned int end, cur;
+		void *dst, *src;
+
+		bvi = container_of(p, struct z_erofs_bvec_item, list);
+		cur = bvi->bvec.offset < 0 ? -bvi->bvec.offset : 0;
+		end = min_t(unsigned int, be->pcl->length - bvi->bvec.offset,
+			    bvi->bvec.end);
+		dst = kmap_local_page(bvi->bvec.page);
+		while (cur < end) {
+			unsigned int pgnr, scur, len;
+
+			pgnr = (bvi->bvec.offset + cur + off0) >> PAGE_SHIFT;
+			DBG_BUGON(pgnr >= be->nr_pages);
+
+			scur = bvi->bvec.offset + cur -
+					((pgnr << PAGE_SHIFT) - off0);
+			len = min_t(unsigned int, end - cur, PAGE_SIZE - scur);
+			if (!be->decompressed_pages[pgnr]) {
+				err = -EFSCORRUPTED;
+				cur += len;
+				continue;
+			}
+			src = kmap_local_page(be->decompressed_pages[pgnr]);
+			memcpy(dst + cur, src + scur, len);
+			kunmap_local(src);
+			cur += len;
+		}
+		kunmap_local(dst);
+		erofs_onlinefolio_end(page_folio(bvi->bvec.page), err, true);
+		list_del(p);
+		kfree(bvi);
+	}
+}
+
+static void z_erofs_parse_out_bvecs(struct z_erofs_backend *be)
+{
+	struct z_erofs_pcluster *pcl = be->pcl;
+	struct z_erofs_bvec_iter biter;
+	struct page *old_bvpage;
+	int i;
+
+	z_erofs_bvec_iter_begin(&biter, &pcl->bvset, Z_EROFS_INLINE_BVECS, 0);
+	for (i = 0; i < pcl->vcnt; ++i) {
+		struct z_erofs_bvec bvec;
+
+		z_erofs_bvec_dequeue(&biter, &bvec, &old_bvpage);
+
+		if (old_bvpage)
+			z_erofs_put_shortlivedpage(be->pagepool, old_bvpage);
+
+		DBG_BUGON(z_erofs_page_is_invalidated(bvec.page));
+		z_erofs_do_decompressed_bvec(be, &bvec);
+	}
+
+	old_bvpage = z_erofs_bvec_iter_end(&biter);
+	if (old_bvpage)
+		z_erofs_put_shortlivedpage(be->pagepool, old_bvpage);
+}
+
+static int z_erofs_parse_in_bvecs(struct z_erofs_backend *be, bool *overlapped)
+{
+	struct z_erofs_pcluster *pcl = be->pcl;
+	unsigned int pclusterpages = z_erofs_pclusterpages(pcl);
+	int i, err = 0;
+
+	*overlapped = false;
+	for (i = 0; i < pclusterpages; ++i) {
+		struct z_erofs_bvec *bvec = &pcl->compressed_bvecs[i];
+		struct page *page = bvec->page;
+
+		/* compressed data ought to be valid when decompressing */
+		if (IS_ERR(page) || !page) {
+			bvec->page = NULL;	/* clear the failure reason */
+			err = page ? PTR_ERR(page) : -EIO;
+			continue;
+		}
+		be->compressed_pages[i] = page;
+
+		if (pcl->from_meta ||
+		    erofs_folio_is_managed(EROFS_SB(be->sb), page_folio(page))) {
+			if (!PageUptodate(page))
+				err = -EIO;
+			continue;
+		}
+
+		DBG_BUGON(z_erofs_page_is_invalidated(page));
+		if (z_erofs_is_shortlived_page(page))
+			continue;
+		z_erofs_do_decompressed_bvec(be, bvec);
+		*overlapped = true;
+	}
+	return err;
+}
+
+static int z_erofs_decompress_pcluster(struct z_erofs_backend *be, bool eio)
+{
+	struct erofs_sb_info *const sbi = EROFS_SB(be->sb);
+	struct z_erofs_pcluster *pcl = be->pcl;
+	unsigned int pclusterpages = z_erofs_pclusterpages(pcl);
+	const struct z_erofs_decompressor *alg =
+				z_erofs_decomp[pcl->algorithmformat];
+	bool try_free = true;
+	int i, j, jtop, err2, err = eio ? -EIO : 0;
+	struct page *page;
+	bool overlapped;
+	const char *reason;
+
+	mutex_lock(&pcl->lock);
+	be->nr_pages = PAGE_ALIGN(pcl->length + pcl->pageofs_out) >> PAGE_SHIFT;
+
+	/* allocate (de)compressed page arrays if cannot be kept on stack */
+	be->decompressed_pages = NULL;
+	be->compressed_pages = NULL;
+	be->onstack_used = 0;
+	if (be->nr_pages <= Z_EROFS_ONSTACK_PAGES) {
+		be->decompressed_pages = be->onstack_pages;
+		be->onstack_used = be->nr_pages;
+		memset(be->decompressed_pages, 0,
+		       sizeof(struct page *) * be->nr_pages);
+	}
+
+	if (pclusterpages + be->onstack_used <= Z_EROFS_ONSTACK_PAGES)
+		be->compressed_pages = be->onstack_pages + be->onstack_used;
+
+	if (!be->decompressed_pages)
+		be->decompressed_pages =
+			kvzalloc_objs(struct page *, be->nr_pages,
+				      GFP_KERNEL | __GFP_NOFAIL);
+	if (!be->compressed_pages)
+		be->compressed_pages =
+			kvzalloc_objs(struct page *, pclusterpages,
+				      GFP_KERNEL | __GFP_NOFAIL);
+
+	z_erofs_parse_out_bvecs(be);
+	err2 = z_erofs_parse_in_bvecs(be, &overlapped);
+	if (err2)
+		err = err2;
+	if (!err) {
+		reason = alg->decompress(&(struct z_erofs_decompress_req) {
+					.sb = be->sb,
+					.in = be->compressed_pages,
+					.out = be->decompressed_pages,
+					.inpages = pclusterpages,
+					.outpages = be->nr_pages,
+					.pageofs_in = pcl->pageofs_in,
+					.pageofs_out = pcl->pageofs_out,
+					.inputsize = pcl->pclustersize,
+					.outputsize = pcl->length,
+					.alg = pcl->algorithmformat,
+					.inplace_io = overlapped,
+					.partial_decoding = pcl->partial,
+					.fillgaps = be->keepxcpy,
+					.gfp = pcl->besteffort ? GFP_KERNEL :
+						GFP_NOWAIT | __GFP_NORETRY
+				 }, be->pagepool);
+		if (IS_ERR(reason)) {
+			if (pcl->besteffort || reason != ERR_PTR(-ENOMEM))
+				erofs_err(be->sb, "failed to decompress (%s) %pe @ pa %llu size %u => %u",
+					  alg->name, reason, pcl->pos,
+					  pcl->pclustersize, pcl->length);
+			err = PTR_ERR(reason);
+		} else if (unlikely(reason)) {
+			erofs_err(be->sb, "failed to decompress (%s) %s @ pa %llu size %u => %u",
+				  alg->name, reason, pcl->pos,
+				  pcl->pclustersize, pcl->length);
+			err = -EFSCORRUPTED;
+		}
+	}
+
+	/* must handle all compressed pages before actual file pages */
+	if (pcl->from_meta) {
+		folio_put(page_folio(pcl->compressed_bvecs[0].page));
+		WRITE_ONCE(pcl->compressed_bvecs[0].page, NULL);
+	} else {
+		/* managed folios are still left in compressed_bvecs[] */
+		for (i = 0; i < pclusterpages; ++i) {
+			page = be->compressed_pages[i];
+			if (!page)
+				continue;
+			if (erofs_folio_is_managed(sbi, page_folio(page))) {
+				try_free = false;
+				continue;
+			}
+			(void)z_erofs_put_shortlivedpage(be->pagepool, page);
+			WRITE_ONCE(pcl->compressed_bvecs[i].page, NULL);
+		}
+	}
+	if (be->compressed_pages < be->onstack_pages ||
+	    be->compressed_pages >= be->onstack_pages + Z_EROFS_ONSTACK_PAGES)
+		kvfree(be->compressed_pages);
+
+	jtop = 0;
+	z_erofs_fill_other_copies(be, err);
+	for (i = 0; i < be->nr_pages; ++i) {
+		page = be->decompressed_pages[i];
+		if (!page)
+			continue;
+
+		DBG_BUGON(z_erofs_page_is_invalidated(page));
+		if (!z_erofs_is_shortlived_page(page)) {
+			erofs_onlinefolio_end(page_folio(page), err, true);
+			continue;
+		}
+		if (pcl->algorithmformat != Z_EROFS_COMPRESSION_LZ4) {
+			erofs_pagepool_add(be->pagepool, page);
+			continue;
+		}
+		for (j = 0; j < jtop && be->decompressed_pages[j] != page; ++j)
+			;
+		if (j >= jtop)	/* this bounce page is newly detected */
+			be->decompressed_pages[jtop++] = page;
+	}
+	while (jtop)
+		erofs_pagepool_add(be->pagepool,
+				   be->decompressed_pages[--jtop]);
+	if (be->decompressed_pages != be->onstack_pages)
+		kvfree(be->decompressed_pages);
+
+	pcl->length = 0;
+	pcl->partial = true;
+	pcl->besteffort = false;
+	pcl->bvset.nextpage = NULL;
+	pcl->vcnt = 0;
+
+	/* pcluster lock MUST be taken before the following line */
+	WRITE_ONCE(pcl->next, NULL);
+	mutex_unlock(&pcl->lock);
+
+	if (pcl->from_meta)
+		z_erofs_free_pcluster(pcl);
+	else
+		z_erofs_put_pcluster(sbi, pcl, try_free);
+	return err;
+}
+
+static int z_erofs_decompress_queue(const struct z_erofs_decompressqueue *io,
+				    struct page **pagepool)
+{
+	struct z_erofs_backend be = {
+		.sb = io->sb,
+		.pagepool = pagepool,
+		.decompressed_secondary_bvecs =
+			LIST_HEAD_INIT(be.decompressed_secondary_bvecs),
+		.pcl = io->head,
+	};
+	struct z_erofs_pcluster *next;
+	int err = 0;
+
+	for (; be.pcl != Z_EROFS_PCLUSTER_TAIL; be.pcl = next) {
+		DBG_BUGON(!be.pcl);
+		next = READ_ONCE(be.pcl->next);
+		err = z_erofs_decompress_pcluster(&be, io->eio) ?: err;
+	}
+	return err;
+}
+
+static void z_erofs_decompressqueue_work(struct work_struct *work)
+{
+	struct z_erofs_decompressqueue *bgq =
+		container_of(work, struct z_erofs_decompressqueue, u.work);
+	struct page *pagepool = NULL;
+
+	DBG_BUGON(bgq->head == Z_EROFS_PCLUSTER_TAIL);
+	z_erofs_decompress_queue(bgq, &pagepool);
+	erofs_release_pages(&pagepool);
+	kvfree(bgq);
+}
+
+#ifdef CONFIG_EROFS_FS_PCPU_KTHREAD
+static void z_erofs_decompressqueue_kthread_work(struct kthread_work *work)
+{
+	z_erofs_decompressqueue_work((struct work_struct *)work);
+}
+#endif
+
+
+static void z_erofs_decompress_kickoff(struct z_erofs_decompressqueue *io,
+				       int bios)
+{
+	struct erofs_sb_info *const sbi = EROFS_SB(io->sb);
+	int gfp_flag;
+
+	/* wake up the caller thread for sync decompression */
+	if (io->sync) {
+		if (!atomic_add_return(bios, &io->pending_bios))
+			complete(&io->u.done);
+		return;
+	}
+
+	if (atomic_add_return(bios, &io->pending_bios))
+		return;
+	if (bio_in_atomic()) {
+		/* See `sync_decompress` in sysfs-fs-erofs for more details */
+		if (sbi->sync_decompress == EROFS_SYNC_DECOMPRESS_AUTO)
+			sbi->sync_decompress = EROFS_SYNC_DECOMPRESS_FORCE_ON;
+#ifdef CONFIG_EROFS_FS_PCPU_KTHREAD
+		scoped_guard(rcu) {
+			struct kthread_worker *worker;
+
+			worker = rcu_dereference(
+				z_erofs_pcpu_workers[raw_smp_processor_id()]);
+			if (worker) {
+				kthread_queue_work(worker, &io->u.kthread_work);
+				return;
+			}
+		}
+		INIT_WORK(&io->u.work, z_erofs_decompressqueue_work);
+#endif
+		queue_work(z_erofs_workqueue, &io->u.work);
+		return;
+	}
+	gfp_flag = memalloc_noio_save();
+	z_erofs_decompressqueue_work(&io->u.work);
+	memalloc_noio_restore(gfp_flag);
+}
+
+static void z_erofs_fill_bio_vec(struct bio_vec *bvec,
+				 struct z_erofs_frontend *f,
+				 struct z_erofs_pcluster *pcl,
+				 unsigned int nr,
+				 struct address_space *mc)
+{
+	gfp_t gfp = mapping_gfp_mask(mc);
+	bool tocache = false;
+	struct z_erofs_bvec zbv;
+	struct address_space *mapping;
+	struct folio *folio;
+	struct page *page;
+	int bs = i_blocksize(f->inode);
+
+	/* Except for inplace folios, the entire folio can be used for I/Os */
+	bvec->bv_offset = 0;
+	bvec->bv_len = PAGE_SIZE;
+repeat:
+	spin_lock(&pcl->lockref.lock);
+	zbv = pcl->compressed_bvecs[nr];
+	spin_unlock(&pcl->lockref.lock);
+	if (!zbv.page)
+		goto out_allocfolio;
+
+	bvec->bv_page = zbv.page;
+	DBG_BUGON(z_erofs_is_shortlived_page(bvec->bv_page));
+
+	folio = page_folio(zbv.page);
+	/*
+	 * Preallocated folios are added to the managed cache here rather than
+	 * in z_erofs_bind_cache() in order to keep these folios locked in
+	 * increasing (physical) address order.
+	 * Clear folio->private before these folios become visible to others in
+	 * the managed cache to avoid duplicate additions for unaligned extents.
+	 */
+	if (folio->private == Z_EROFS_PREALLOCATED_FOLIO) {
+		folio->private = NULL;
+		tocache = true;
+		goto out_tocache;
+	}
+
+	mapping = READ_ONCE(folio->mapping);
+	/*
+	 * File-backed folios for inplace I/Os are all locked steady,
+	 * therefore it is impossible for `mapping` to be NULL.
+	 */
+	if (mapping && mapping != mc) {
+		if (zbv.offset < 0)
+			bvec->bv_offset = round_up(-zbv.offset, bs);
+		bvec->bv_len = round_up(zbv.end, bs) - bvec->bv_offset;
+		return;
+	}
+
+	folio_lock(folio);
+	if (likely(folio->mapping == mc)) {
+		/*
+		 * The cached folio is still in managed cache but without
+		 * a valid `->private` pcluster hint.  Let's reconnect them.
+		 */
+		if (!folio_test_private(folio)) {
+			folio_attach_private(folio, pcl);
+			/* compressed_bvecs[] already takes a ref before */
+			folio_put(folio);
+		}
+		if (likely(folio->private == pcl))  {
+			/* don't submit cache I/Os again if already uptodate */
+			if (folio_test_uptodate(folio)) {
+				folio_unlock(folio);
+				bvec->bv_page = NULL;
+			}
+			return;
+		}
+	} else {
+		DBG_BUGON(1); /* referenced managed folios can't be truncated */
+	}
+	folio_unlock(folio);
+	folio_put(folio);
+out_allocfolio:
+	page = __erofs_allocpage(&f->pagepool, gfp, true);
+	spin_lock(&pcl->lockref.lock);
+	if (unlikely(pcl->compressed_bvecs[nr].page != zbv.page)) {
+		if (page)
+			erofs_pagepool_add(&f->pagepool, page);
+		spin_unlock(&pcl->lockref.lock);
+		cond_resched();
+		goto repeat;
+	}
+	pcl->compressed_bvecs[nr].page = page ? page : ERR_PTR(-ENOMEM);
+	spin_unlock(&pcl->lockref.lock);
+	bvec->bv_page = page;
+	if (!page)
+		return;
+	folio = page_folio(page);
+out_tocache:
+	if (!tocache || bs != PAGE_SIZE ||
+	    filemap_add_folio(mc, folio, (pcl->pos >> PAGE_SHIFT) + nr, gfp)) {
+		/* turn into a temporary shortlived folio (1 ref) */
+		folio->private = (void *)Z_EROFS_SHORTLIVED_PAGE;
+		return;
+	}
+	folio_attach_private(folio, pcl);
+	/* drop a refcount added by allocpage (then 2 refs in total here) */
+	folio_put(folio);
+}
+
+static struct z_erofs_decompressqueue *jobqueue_init(struct super_block *sb,
+			      struct z_erofs_decompressqueue *fgq, bool *fg)
+{
+	struct z_erofs_decompressqueue *q;
+
+	if (fg && !*fg) {
+		q = kvzalloc_obj(*q, GFP_KERNEL | __GFP_NOWARN);
+		if (!q) {
+			*fg = true;
+			goto fg_out;
+		}
+#ifdef CONFIG_EROFS_FS_PCPU_KTHREAD
+		kthread_init_work(&q->u.kthread_work,
+				  z_erofs_decompressqueue_kthread_work);
+#else
+		INIT_WORK(&q->u.work, z_erofs_decompressqueue_work);
+#endif
+	} else {
+fg_out:
+		q = fgq;
+		init_completion(&fgq->u.done);
+		atomic_set(&fgq->pending_bios, 0);
+		q->eio = false;
+		q->sync = true;
+	}
+	q->sb = sb;
+	q->head = Z_EROFS_PCLUSTER_TAIL;
+	return q;
+}
+
+/* define decompression jobqueue types */
+enum {
+	JQ_BYPASS,
+	JQ_SUBMIT,
+	NR_JOBQUEUES,
+};
+
+static void z_erofs_move_to_bypass_queue(struct z_erofs_pcluster *pcl,
+					 struct z_erofs_pcluster *next,
+					 struct z_erofs_pcluster **qtail[])
+{
+	WRITE_ONCE(pcl->next, Z_EROFS_PCLUSTER_TAIL);
+	WRITE_ONCE(*qtail[JQ_SUBMIT], next);
+	WRITE_ONCE(*qtail[JQ_BYPASS], pcl);
+	qtail[JQ_BYPASS] = &pcl->next;
+}
+
+static void z_erofs_endio(struct bio *bio)
+{
+	struct z_erofs_decompressqueue *q = bio->bi_private;
+	blk_status_t err = bio->bi_status;
+	struct folio_iter fi;
+
+	bio_for_each_folio_all(fi, bio) {
+		struct folio *folio = fi.folio;
+
+		DBG_BUGON(folio_test_uptodate(folio));
+		DBG_BUGON(z_erofs_page_is_invalidated(&folio->page));
+		if (!erofs_folio_is_managed(EROFS_SB(q->sb), folio))
+			continue;
+
+		if (!err)
+			folio_mark_uptodate(folio);
+		folio_unlock(folio);
+	}
+	if (err)
+		q->eio = true;
+	z_erofs_decompress_kickoff(q, -1);
+	if (bio->bi_bdev)
+		bio_put(bio);
+}
+
+static void z_erofs_submit_queue(struct z_erofs_frontend *f,
+				 struct z_erofs_decompressqueue *fgq,
+				 bool *force_fg, bool readahead)
+{
+	struct super_block *sb = f->inode->i_sb;
+	struct address_space *mc = MNGD_MAPPING(EROFS_SB(sb));
+	struct z_erofs_pcluster **qtail[NR_JOBQUEUES];
+	struct z_erofs_decompressqueue *q[NR_JOBQUEUES];
+	struct z_erofs_pcluster *pcl, *next;
+	/* bio is NULL initially, so no need to initialize last_{index,bdev} */
+	erofs_off_t last_pa;
+	unsigned int nr_bios = 0;
+	struct bio *bio = NULL;
+	unsigned long pflags;
+	int memstall = 0;
+
+	/* No need to read from device for pclusters in the bypass queue. */
+	q[JQ_BYPASS] = jobqueue_init(sb, fgq + JQ_BYPASS, NULL);
+	q[JQ_SUBMIT] = jobqueue_init(sb, fgq + JQ_SUBMIT, force_fg);
+
+	qtail[JQ_BYPASS] = &q[JQ_BYPASS]->head;
+	qtail[JQ_SUBMIT] = &q[JQ_SUBMIT]->head;
+
+	/* by default, all need io submission */
+	q[JQ_SUBMIT]->head = next = f->head;
+
+	do {
+		struct erofs_map_dev mdev;
+		erofs_off_t cur, end;
+		struct bio_vec bvec;
+		unsigned int i = 0;
+		bool bypass = true;
+
+		pcl = next;
+		next = READ_ONCE(pcl->next);
+		if (pcl->from_meta) {
+			z_erofs_move_to_bypass_queue(pcl, next, qtail);
+			continue;
+		}
+
+		/* no device id here, thus it will always succeed */
+		mdev = (struct erofs_map_dev) {
+			.m_pa = round_down(pcl->pos, sb->s_blocksize),
+		};
+		(void)erofs_map_dev(sb, &mdev);
+
+		cur = mdev.m_pa;
+		end = round_up(cur + pcl->pageofs_in + pcl->pclustersize,
+			       sb->s_blocksize);
+		do {
+			bvec.bv_page = NULL;
+			if (bio && (cur != last_pa ||
+				    bio->bi_bdev != mdev.m_bdev)) {
+drain_io:
+				if (erofs_is_fileio_mode(EROFS_SB(sb)))
+					erofs_fileio_submit_bio(bio);
+				else
+					submit_bio(bio);
+
+				if (memstall) {
+					psi_memstall_leave(&pflags);
+					memstall = 0;
+				}
+				bio = NULL;
+			}
+
+			if (!bvec.bv_page) {
+				z_erofs_fill_bio_vec(&bvec, f, pcl, i++, mc);
+				if (!bvec.bv_page)
+					continue;
+				if (cur + bvec.bv_len > end)
+					bvec.bv_len = end - cur;
+				DBG_BUGON(bvec.bv_len < sb->s_blocksize);
+			}
+
+			if (unlikely(PageWorkingset(bvec.bv_page)) &&
+			    !memstall) {
+				psi_memstall_enter(&pflags);
+				memstall = 1;
+			}
+
+			if (!bio) {
+				if (erofs_is_fileio_mode(EROFS_SB(sb)))
+					bio = erofs_fileio_bio_alloc(&mdev);
+				else
+					bio = bio_alloc(mdev.m_bdev, BIO_MAX_VECS,
+							REQ_OP_READ, GFP_NOIO);
+				bio->bi_end_io = z_erofs_endio;
+				bio->bi_iter.bi_sector =
+						(mdev.m_dif->fsoff + cur) >> 9;
+				bio->bi_private = q[JQ_SUBMIT];
+				if (readahead)
+					bio->bi_opf |= REQ_RAHEAD;
+				++nr_bios;
+			}
+
+			if (!bio_add_page(bio, bvec.bv_page, bvec.bv_len,
+					  bvec.bv_offset))
+				goto drain_io;
+			last_pa = cur + bvec.bv_len;
+			bypass = false;
+		} while ((cur += bvec.bv_len) < end);
+
+		if (!bypass)
+			qtail[JQ_SUBMIT] = &pcl->next;
+		else
+			z_erofs_move_to_bypass_queue(pcl, next, qtail);
+	} while (next != Z_EROFS_PCLUSTER_TAIL);
+
+	if (bio) {
+		if (erofs_is_fileio_mode(EROFS_SB(sb)))
+			erofs_fileio_submit_bio(bio);
+		else
+			submit_bio(bio);
+	}
+	if (memstall)
+		psi_memstall_leave(&pflags);
+
+	/*
+	 * although background is preferred, no one is pending for submission.
+	 * don't issue decompression but drop it directly instead.
+	 */
+	if (!*force_fg && !nr_bios) {
+		kvfree(q[JQ_SUBMIT]);
+		return;
+	}
+	z_erofs_decompress_kickoff(q[JQ_SUBMIT], nr_bios);
+}
+
+static int z_erofs_runqueue(struct z_erofs_frontend *f, unsigned int rabytes)
+{
+	struct z_erofs_decompressqueue io[NR_JOBQUEUES];
+	struct erofs_sb_info *sbi = EROFS_I_SB(f->inode);
+	int syncmode = sbi->sync_decompress;
+	bool force_fg;
+	int err;
+
+	force_fg = (syncmode == EROFS_SYNC_DECOMPRESS_AUTO && !rabytes) ||
+		(syncmode == EROFS_SYNC_DECOMPRESS_FORCE_ON &&
+			(rabytes <= Z_EROFS_MAX_SYNC_DECOMPRESS_BYTES));
+
+	if (f->head == Z_EROFS_PCLUSTER_TAIL)
+		return 0;
+	z_erofs_submit_queue(f, io, &force_fg, !!rabytes);
+
+	/* handle bypass queue (no i/o pclusters) immediately */
+	err = z_erofs_decompress_queue(&io[JQ_BYPASS], &f->pagepool);
+	if (!force_fg)
+		return err;
+
+	/* wait until all bios are completed */
+	wait_for_completion_io(&io[JQ_SUBMIT].u.done);
+
+	/* handle synchronous decompress queue in the caller context */
+	return z_erofs_decompress_queue(&io[JQ_SUBMIT], &f->pagepool) ?: err;
+}
+
+/*
+ * Since partial uptodate is still unimplemented for now, we have to use
+ * approximate readmore strategies as a start.
+ */
+static void z_erofs_pcluster_readmore(struct z_erofs_frontend *f,
+		struct readahead_control *rac, bool backmost)
+{
+	struct inode *inode = f->inode;
+	struct erofs_map_blocks *map = &f->map;
+	erofs_off_t cur, end, headoffset = f->headoffset;
+	int err;
+
+	if (backmost) {
+		if (rac)
+			end = headoffset + readahead_length(rac) - 1;
+		else
+			end = headoffset + PAGE_SIZE - 1;
+		map->m_la = end;
+		err = z_erofs_map_blocks_iter(inode, map,
+					      EROFS_GET_BLOCKS_READMORE);
+		if (err || !(map->m_flags & EROFS_MAP_MAPPED))
+			return;
+
+		/* expand ra for the trailing edge if readahead */
+		if (rac) {
+			cur = round_up(map->m_la + map->m_llen, PAGE_SIZE);
+			readahead_expand(rac, headoffset, cur - headoffset);
+			return;
+		}
+		end = round_up(end, PAGE_SIZE);
+	} else {
+		end = round_up(map->m_la, PAGE_SIZE);
+		if (!(map->m_flags & EROFS_MAP_MAPPED) || !map->m_llen)
+			return;
+	}
+
+	cur = map->m_la + map->m_llen - 1;
+	while ((cur >= end) && (cur < i_size_read(inode))) {
+		pgoff_t index = cur >> PAGE_SHIFT;
+		struct folio *folio;
+
+		folio = erofs_grab_folio_nowait(f->sharedinode->i_mapping, index);
+		if (!IS_ERR_OR_NULL(folio)) {
+			if (folio_test_uptodate(folio))
+				folio_unlock(folio);
+			else
+				z_erofs_scan_folio(f, folio, !!rac);
+			folio_put(folio);
+		}
+
+		if (cur < PAGE_SIZE)
+			break;
+		cur = ((loff_t)index << PAGE_SHIFT) - 1;
+	}
+}
+
+static int z_erofs_read_folio(struct file *file, struct folio *folio)
+{
+	struct inode *sharedinode = folio->mapping->host;
+	bool need_iput;
+	struct inode *realinode = erofs_real_inode(sharedinode, &need_iput);
+	Z_EROFS_DEFINE_FRONTEND(f, realinode, sharedinode, folio_pos(folio));
+	int err;
+
+	trace_erofs_read_folio(realinode, folio, false);
+	z_erofs_pcluster_readmore(&f, NULL, true);
+	err = z_erofs_scan_folio(&f, folio, false);
+	z_erofs_pcluster_readmore(&f, NULL, false);
+	z_erofs_pcluster_end(&f);
+
+	/* if some pclusters are ready, need submit them anyway */
+	err = z_erofs_runqueue(&f, 0) ?: err;
+	if (err && err != -EINTR)
+		erofs_err(realinode->i_sb, "read error %d @ %lu of nid %llu",
+			  err, folio->index, EROFS_I(realinode)->nid);
+
+	erofs_put_metabuf(&f.map.buf);
+	erofs_release_pages(&f.pagepool);
+
+	if (need_iput)
+		iput(realinode);
+	return err;
+}
+
+static void z_erofs_readahead(struct readahead_control *rac)
+{
+	struct inode *sharedinode = rac->mapping->host;
+	bool need_iput;
+	struct inode *realinode = erofs_real_inode(sharedinode, &need_iput);
+	Z_EROFS_DEFINE_FRONTEND(f, realinode, sharedinode, readahead_pos(rac));
+	unsigned int nrpages = readahead_count(rac);
+	struct folio *head = NULL, *folio;
+	int err;
+
+	trace_erofs_readahead(realinode, readahead_index(rac), nrpages, false);
+	z_erofs_pcluster_readmore(&f, rac, true);
+	while ((folio = readahead_folio(rac))) {
+		folio->private = head;
+		head = folio;
+	}
+
+	/* traverse in reverse order for best metadata I/O performance */
+	while (head) {
+		folio = head;
+		head = folio_get_private(folio);
+
+		err = z_erofs_scan_folio(&f, folio, true);
+		if (err && err != -EINTR)
+			erofs_err(realinode->i_sb, "readahead error at folio %lu @ nid %llu",
+				  folio->index, EROFS_I(realinode)->nid);
+	}
+	z_erofs_pcluster_readmore(&f, rac, false);
+	z_erofs_pcluster_end(&f);
+
+	(void)z_erofs_runqueue(&f, nrpages << PAGE_SHIFT);
+	erofs_put_metabuf(&f.map.buf);
+	erofs_release_pages(&f.pagepool);
+
+	if (need_iput)
+		iput(realinode);
+}
+
+const struct address_space_operations z_erofs_aops = {
+	.read_folio = z_erofs_read_folio,
+	.readahead = z_erofs_readahead,
+};
diff --git a/fs/erofs/zmap.c b/fs/erofs/zmap.c
new file mode 100644
index 000000000..f8981fc74
--- /dev/null
+++ b/fs/erofs/zmap.c
@@ -0,0 +1,838 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2018-2019 HUAWEI, Inc.
+ *             https://www.huawei.com/
+ */
+#include "internal.h"
+#include <linux/unaligned.h>
+#include <trace/events/erofs.h>
+
+struct z_erofs_maprecorder {
+	struct inode *inode;
+	struct erofs_map_blocks *map;
+	u64 lcn;
+	/* compression extent information gathered */
+	u8  type, headtype;
+	u16 clusterofs;
+	u16 delta[2];
+	erofs_blk_t pblk;
+	erofs_off_t nextpackoff;
+	int compressedblks;
+	bool partialref, in_mbox;
+};
+
+static int z_erofs_load_full_lcluster(struct z_erofs_maprecorder *m, u64 lcn)
+{
+	struct inode *const inode = m->inode;
+	struct erofs_inode *const vi = EROFS_I(inode);
+	const erofs_off_t pos = Z_EROFS_FULL_INDEX_START(erofs_iloc(inode) +
+			vi->inode_isize + vi->xattr_isize) +
+			lcn * sizeof(struct z_erofs_lcluster_index);
+	struct z_erofs_lcluster_index *di;
+	unsigned int advise;
+
+	di = erofs_read_metabuf(&m->map->buf, inode->i_sb, pos, m->in_mbox);
+	if (IS_ERR(di))
+		return PTR_ERR(di);
+	m->lcn = lcn;
+	m->nextpackoff = pos + sizeof(struct z_erofs_lcluster_index);
+
+	advise = le16_to_cpu(di->di_advise);
+	m->type = advise & Z_EROFS_LI_LCLUSTER_TYPE_MASK;
+	if (m->type == Z_EROFS_LCLUSTER_TYPE_NONHEAD) {
+		m->clusterofs = 1 << vi->z_lclusterbits;
+		m->delta[0] = le16_to_cpu(di->di_u.delta[0]);
+		if (m->delta[0] & Z_EROFS_LI_D0_CBLKCNT) {
+			if (!(vi->z_advise & (Z_EROFS_ADVISE_BIG_PCLUSTER_1 |
+					Z_EROFS_ADVISE_BIG_PCLUSTER_2))) {
+				DBG_BUGON(1);
+				return -EFSCORRUPTED;
+			}
+			m->compressedblks = m->delta[0] & ~Z_EROFS_LI_D0_CBLKCNT;
+			m->delta[0] = 1;
+		}
+		m->delta[1] = le16_to_cpu(di->di_u.delta[1]);
+	} else {
+		m->partialref = !!(advise & Z_EROFS_LI_PARTIAL_REF);
+		m->clusterofs = le16_to_cpu(di->di_clusterofs);
+		if (advise & Z_EROFS_LI_HOLE) {
+			m->compressedblks = 0;
+			m->pblk = EROFS_NULL_ADDR;
+		} else {
+			m->pblk = le32_to_cpu(di->di_u.blkaddr);
+		}
+	}
+	return 0;
+}
+
+static unsigned int decode_compactedbits(unsigned int lobits,
+					 u8 *in, unsigned int pos, u8 *type)
+{
+	const unsigned int v = get_unaligned_le32(in + pos / 8) >> (pos & 7);
+	const unsigned int lo = v & ((1 << lobits) - 1);
+
+	*type = (v >> lobits) & 3;
+	return lo;
+}
+
+static int get_compacted_la_distance(unsigned int lobits,
+				     unsigned int encodebits,
+				     unsigned int vcnt, u8 *in, int i)
+{
+	unsigned int lo, d1 = 0;
+	u8 type;
+
+	DBG_BUGON(i >= vcnt);
+
+	do {
+		lo = decode_compactedbits(lobits, in, encodebits * i, &type);
+
+		if (type != Z_EROFS_LCLUSTER_TYPE_NONHEAD)
+			return d1;
+		++d1;
+	} while (++i < vcnt);
+
+	/* vcnt - 1 (Z_EROFS_LCLUSTER_TYPE_NONHEAD) item */
+	if (!(lo & Z_EROFS_LI_D0_CBLKCNT))
+		d1 += lo - 1;
+	return d1;
+}
+
+static int z_erofs_load_compact_lcluster(struct z_erofs_maprecorder *m,
+					 u64 lcn, bool lookahead)
+{
+	struct inode *const inode = m->inode;
+	struct erofs_inode *const vi = EROFS_I(inode);
+	const erofs_off_t ebase = Z_EROFS_MAP_HEADER_END(erofs_iloc(inode) +
+			vi->inode_isize + vi->xattr_isize);
+	const unsigned int lclusterbits = vi->z_lclusterbits;
+	const unsigned int totalidx = erofs_iblks(inode);
+	unsigned int compacted_4b_initial, compacted_2b, amortizedshift;
+	unsigned int vcnt, lo, lobits, encodebits, nblk, bytes;
+	bool big_pcluster = vi->z_advise & Z_EROFS_ADVISE_BIG_PCLUSTER_1;
+	erofs_off_t pos;
+	u8 *in, type;
+	int i;
+
+	if (lcn >= totalidx || lclusterbits > 14)
+		return -EINVAL;
+
+	m->lcn = lcn;
+	/* used to align to 32-byte (compacted_2b) alignment */
+	compacted_4b_initial = ((32 - ebase % 32) / 4) & 7;
+	compacted_2b = 0;
+	if ((vi->z_advise & Z_EROFS_ADVISE_COMPACTED_2B) &&
+	    compacted_4b_initial < totalidx)
+		compacted_2b = rounddown(totalidx - compacted_4b_initial, 16);
+
+	pos = ebase;
+	amortizedshift = 2;	/* compact_4b */
+	if (lcn >= compacted_4b_initial) {
+		pos += compacted_4b_initial * 4;
+		lcn -= compacted_4b_initial;
+		if (lcn < compacted_2b) {
+			amortizedshift = 1;
+		} else {
+			pos += compacted_2b * 2;
+			lcn -= compacted_2b;
+		}
+	}
+	pos += lcn * (1 << amortizedshift);
+
+	/* figure out the lcluster count in this pack */
+	if (1 << amortizedshift == 4 && lclusterbits <= 14)
+		vcnt = 2;
+	else if (1 << amortizedshift == 2 && lclusterbits <= 12)
+		vcnt = 16;
+	else
+		return -EOPNOTSUPP;
+
+	in = erofs_read_metabuf(&m->map->buf, inode->i_sb, pos, m->in_mbox);
+	if (IS_ERR(in))
+		return PTR_ERR(in);
+
+	/* it doesn't equal to round_up(..) */
+	m->nextpackoff = round_down(pos, vcnt << amortizedshift) +
+			 (vcnt << amortizedshift);
+	lobits = max(lclusterbits, ilog2(Z_EROFS_LI_D0_CBLKCNT) + 1U);
+	encodebits = ((vcnt << amortizedshift) - sizeof(__le32)) * 8 / vcnt;
+	bytes = pos & ((vcnt << amortizedshift) - 1);
+	in -= bytes;
+	i = bytes >> amortizedshift;
+
+	lo = decode_compactedbits(lobits, in, encodebits * i, &type);
+	m->type = type;
+	if (type == Z_EROFS_LCLUSTER_TYPE_NONHEAD) {
+		m->clusterofs = 1 << lclusterbits;
+
+		/* figure out lookahead_distance: delta[1] if needed */
+		if (lookahead)
+			m->delta[1] = get_compacted_la_distance(lobits,
+						encodebits, vcnt, in, i);
+		if (lo & Z_EROFS_LI_D0_CBLKCNT) {
+			if (!big_pcluster) {
+				DBG_BUGON(1);
+				return -EFSCORRUPTED;
+			}
+			m->compressedblks = lo & ~Z_EROFS_LI_D0_CBLKCNT;
+			m->delta[0] = 1;
+			return 0;
+		} else if (i + 1 != (int)vcnt) {
+			m->delta[0] = lo;
+			return 0;
+		}
+		/*
+		 * since the last lcluster in the pack is special,
+		 * of which lo saves delta[1] rather than delta[0].
+		 * Hence, get delta[0] by the previous lcluster indirectly.
+		 */
+		lo = decode_compactedbits(lobits, in,
+					  encodebits * (i - 1), &type);
+		if (type != Z_EROFS_LCLUSTER_TYPE_NONHEAD)
+			lo = 0;
+		else if (lo & Z_EROFS_LI_D0_CBLKCNT)
+			lo = 1;
+		m->delta[0] = lo + 1;
+		return 0;
+	}
+	m->clusterofs = lo;
+	m->delta[0] = 0;
+	/* figout out blkaddr (pblk) for HEAD lclusters */
+	if (!big_pcluster) {
+		nblk = 1;
+		while (i > 0) {
+			--i;
+			lo = decode_compactedbits(lobits, in,
+						  encodebits * i, &type);
+			if (type == Z_EROFS_LCLUSTER_TYPE_NONHEAD)
+				i -= lo;
+
+			if (i >= 0)
+				++nblk;
+		}
+	} else {
+		nblk = 0;
+		while (i > 0) {
+			--i;
+			lo = decode_compactedbits(lobits, in,
+						  encodebits * i, &type);
+			if (type == Z_EROFS_LCLUSTER_TYPE_NONHEAD) {
+				if (lo & Z_EROFS_LI_D0_CBLKCNT) {
+					--i;
+					nblk += lo & ~Z_EROFS_LI_D0_CBLKCNT;
+					continue;
+				}
+				/* bigpcluster shouldn't have plain d0 == 1 */
+				if (lo <= 1) {
+					DBG_BUGON(1);
+					return -EFSCORRUPTED;
+				}
+				i -= lo - 2;
+				continue;
+			}
+			++nblk;
+		}
+	}
+	in += (vcnt << amortizedshift) - sizeof(__le32);
+	m->pblk = le32_to_cpu(*(__le32 *)in) + nblk;
+	return 0;
+}
+
+static int z_erofs_load_lcluster_from_disk(struct z_erofs_maprecorder *m,
+					   u64 lcn, bool lookahead)
+{
+	struct erofs_inode *vi = EROFS_I(m->inode);
+	int err;
+
+	if (vi->datalayout == EROFS_INODE_COMPRESSED_COMPACT) {
+		err = z_erofs_load_compact_lcluster(m, lcn, lookahead);
+	} else {
+		DBG_BUGON(vi->datalayout != EROFS_INODE_COMPRESSED_FULL);
+		err = z_erofs_load_full_lcluster(m, lcn);
+	}
+	if (err)
+		return err;
+
+	if (m->type >= Z_EROFS_LCLUSTER_TYPE_MAX) {
+		erofs_err(m->inode->i_sb, "unknown type %u @ lcn %llu of nid %llu",
+			  m->type, lcn, EROFS_I(m->inode)->nid);
+		DBG_BUGON(1);
+		return -EOPNOTSUPP;
+	} else if (m->type != Z_EROFS_LCLUSTER_TYPE_NONHEAD &&
+		   m->clusterofs >= (1 << vi->z_lclusterbits)) {
+		DBG_BUGON(1);
+		return -EFSCORRUPTED;
+	}
+	return 0;
+}
+
+static int z_erofs_extent_lookback(struct z_erofs_maprecorder *m,
+				   unsigned int lookback_distance)
+{
+	struct super_block *sb = m->inode->i_sb;
+	struct erofs_inode *const vi = EROFS_I(m->inode);
+	const unsigned int lclusterbits = vi->z_lclusterbits;
+
+	while (m->lcn >= lookback_distance) {
+		u64 lcn = m->lcn - lookback_distance;
+		int err;
+
+		if (!lookback_distance)
+			break;
+
+		err = z_erofs_load_lcluster_from_disk(m, lcn, false);
+		if (err)
+			return err;
+		if (m->type == Z_EROFS_LCLUSTER_TYPE_NONHEAD) {
+			lookback_distance = m->delta[0];
+			continue;
+		}
+		m->headtype = m->type;
+		m->map->m_la = (lcn << lclusterbits) | m->clusterofs;
+		return 0;
+	}
+	erofs_err(sb, "bogus lookback distance %u @ lcn %llu of nid %llu",
+		  lookback_distance, m->lcn, vi->nid);
+	DBG_BUGON(1);
+	return -EFSCORRUPTED;
+}
+
+static int z_erofs_get_extent_compressedlen(struct z_erofs_maprecorder *m,
+					    unsigned int initial_lcn)
+{
+	struct inode *inode = m->inode;
+	struct super_block *sb = inode->i_sb;
+	struct erofs_inode *vi = EROFS_I(inode);
+	bool bigpcl1 = vi->z_advise & Z_EROFS_ADVISE_BIG_PCLUSTER_1;
+	bool bigpcl2 = vi->z_advise & Z_EROFS_ADVISE_BIG_PCLUSTER_2;
+	u64 lcn = m->lcn + 1;
+	int err;
+
+	DBG_BUGON(m->type == Z_EROFS_LCLUSTER_TYPE_NONHEAD);
+	DBG_BUGON(m->type != m->headtype);
+
+	if ((m->headtype == Z_EROFS_LCLUSTER_TYPE_HEAD1 && !bigpcl1) ||
+	    ((m->headtype == Z_EROFS_LCLUSTER_TYPE_PLAIN ||
+	      m->headtype == Z_EROFS_LCLUSTER_TYPE_HEAD2) && !bigpcl2) ||
+	    (lcn << vi->z_lclusterbits) >= inode->i_size)
+		if (m->compressedblks < 0)
+			m->compressedblks = 1;
+
+	if (m->compressedblks >= 0)
+		goto out;
+
+	err = z_erofs_load_lcluster_from_disk(m, lcn, false);
+	if (err)
+		return err;
+
+	/*
+	 * If the 1st NONHEAD lcluster has already been handled initially w/o
+	 * valid compressedblks, which means at least it mustn't be CBLKCNT, or
+	 * an internal implemenatation error is detected.
+	 *
+	 * The following code can also handle it properly anyway, but let's
+	 * BUG_ON in the debugging mode only for developers to notice that.
+	 */
+	DBG_BUGON(lcn == initial_lcn &&
+		  m->type == Z_EROFS_LCLUSTER_TYPE_NONHEAD);
+
+	if (m->type != Z_EROFS_LCLUSTER_TYPE_NONHEAD) {
+		/*
+		 * if the 1st NONHEAD lcluster is actually PLAIN or HEAD type
+		 * rather than CBLKCNT, it's a 1 block-sized pcluster.
+		 */
+		if (m->compressedblks < 0)
+			m->compressedblks = 1;
+	} else if (m->delta[0] != 1 || m->compressedblks < 0) {
+		erofs_err(sb, "bogus CBLKCNT @ lcn %llu of nid %llu", lcn, vi->nid);
+		DBG_BUGON(1);
+		return -EFSCORRUPTED;
+	}
+
+out:
+	if (!m->compressedblks)
+		m->map->m_flags &= ~EROFS_MAP_MAPPED;
+	m->map->m_plen = erofs_pos(sb, m->compressedblks);
+	return 0;
+}
+
+static int z_erofs_get_extent_decompressedlen(struct z_erofs_maprecorder *m)
+{
+	struct inode *inode = m->inode;
+	struct erofs_inode *vi = EROFS_I(inode);
+	struct erofs_map_blocks *map = m->map;
+	unsigned int lclusterbits = vi->z_lclusterbits;
+	u64 lcn = m->lcn, headlcn = map->m_la >> lclusterbits;
+	int err;
+
+	while (1) {
+		/* handle the last EOF pcluster (no next HEAD lcluster) */
+		if ((lcn << lclusterbits) >= inode->i_size) {
+			map->m_llen = inode->i_size - map->m_la;
+			return 0;
+		}
+
+		err = z_erofs_load_lcluster_from_disk(m, lcn, true);
+		if (err)
+			return err;
+
+		if (m->type == Z_EROFS_LCLUSTER_TYPE_NONHEAD) {
+			/* work around invalid d1 generated by pre-1.0 mkfs */
+			if (unlikely(!m->delta[1])) {
+				m->delta[1] = 1;
+				DBG_BUGON(1);
+			}
+		} else if (m->type < Z_EROFS_LCLUSTER_TYPE_MAX) {
+			if (lcn != headlcn)
+				break;	/* ends at the next HEAD lcluster */
+			m->delta[1] = 1;
+		}
+		lcn += m->delta[1];
+	}
+	map->m_llen = (lcn << lclusterbits) + m->clusterofs - map->m_la;
+	return 0;
+}
+
+static int z_erofs_map_blocks_fo(struct inode *inode,
+				 struct erofs_map_blocks *map, int flags)
+{
+	struct erofs_inode *vi = EROFS_I(inode);
+	struct super_block *sb = inode->i_sb;
+	bool fragment = vi->z_advise & Z_EROFS_ADVISE_FRAGMENT_PCLUSTER;
+	bool ztailpacking = vi->z_idata_size;
+	unsigned int lclusterbits = vi->z_lclusterbits;
+	struct z_erofs_maprecorder m = {
+		.inode = inode,
+		.map = map,
+		.in_mbox = erofs_inode_in_metabox(inode),
+		.compressedblks = -1,
+	};
+	unsigned int endoff;
+	unsigned long initial_lcn;
+	unsigned long long ofs, end;
+	int err;
+
+	ofs = flags & EROFS_GET_BLOCKS_FINDTAIL ? inode->i_size - 1 : map->m_la;
+	if (fragment && !(flags & EROFS_GET_BLOCKS_FINDTAIL) &&
+	    !vi->z_tailextent_headlcn) {
+		map->m_la = 0;
+		map->m_llen = inode->i_size;
+		map->m_flags = EROFS_MAP_FRAGMENT;
+		return 0;
+	}
+	initial_lcn = ofs >> lclusterbits;
+	endoff = ofs & ((1 << lclusterbits) - 1);
+
+	err = z_erofs_load_lcluster_from_disk(&m, initial_lcn, false);
+	if (err)
+		goto unmap_out;
+
+	if ((flags & EROFS_GET_BLOCKS_FINDTAIL) && ztailpacking)
+		vi->z_fragmentoff = m.nextpackoff;
+	map->m_flags = EROFS_MAP_MAPPED | EROFS_MAP_PARTIAL_MAPPED;
+	end = (m.lcn + 1ULL) << lclusterbits;
+
+	if (m.type != Z_EROFS_LCLUSTER_TYPE_NONHEAD && endoff >= m.clusterofs) {
+		m.headtype = m.type;
+		map->m_la = (m.lcn << lclusterbits) | m.clusterofs;
+		/*
+		 * For ztailpacking files, in order to inline data more
+		 * effectively, special EOF lclusters are now supported
+		 * which can have three parts at most.
+		 */
+		if (ztailpacking && end > inode->i_size)
+			end = inode->i_size;
+	} else {
+		if (m.type != Z_EROFS_LCLUSTER_TYPE_NONHEAD) {
+			end = (m.lcn << lclusterbits) | m.clusterofs;
+			map->m_flags &= ~EROFS_MAP_PARTIAL_MAPPED;
+			m.delta[0] = 1;
+		}
+		/* get the corresponding first chunk */
+		err = z_erofs_extent_lookback(&m, m.delta[0]);
+		if (err)
+			goto unmap_out;
+	}
+	if (m.partialref)
+		map->m_flags |= EROFS_MAP_PARTIAL_REF;
+	map->m_llen = end - map->m_la;
+
+	if (flags & EROFS_GET_BLOCKS_FINDTAIL) {
+		vi->z_tailextent_headlcn = m.lcn;
+		/* for non-compact indexes, fragmentoff is 64 bits */
+		if (fragment && vi->datalayout == EROFS_INODE_COMPRESSED_FULL)
+			vi->z_fragmentoff |= (u64)m.pblk << 32;
+	}
+	if (ztailpacking && m.lcn == vi->z_tailextent_headlcn) {
+		map->m_flags |= EROFS_MAP_META;
+		map->m_pa = vi->z_fragmentoff;
+		map->m_plen = vi->z_idata_size;
+		if (erofs_blkoff(sb, map->m_pa) + map->m_plen > sb->s_blocksize) {
+			erofs_err(sb, "ztailpacking inline data across blocks @ nid %llu",
+				  vi->nid);
+			err = -EFSCORRUPTED;
+			goto unmap_out;
+		}
+	} else if (fragment && m.lcn == vi->z_tailextent_headlcn) {
+		map->m_flags = EROFS_MAP_FRAGMENT;
+	} else {
+		map->m_pa = erofs_pos(sb, m.pblk);
+		err = z_erofs_get_extent_compressedlen(&m, initial_lcn);
+		if (err)
+			goto unmap_out;
+	}
+
+	if (m.headtype == Z_EROFS_LCLUSTER_TYPE_PLAIN) {
+		if ((vi->z_advise & Z_EROFS_ADVISE_INTERLACED_PCLUSTER) &&
+		    !(map->m_flags & EROFS_MAP_META))
+			map->m_algorithmformat = Z_EROFS_COMPRESSION_INTERLACED;
+		else
+			map->m_algorithmformat = Z_EROFS_COMPRESSION_SHIFTED;
+	} else {
+		map->m_algorithmformat =
+			vi->z_algofmt[m.headtype == Z_EROFS_LCLUSTER_TYPE_HEAD2];
+	}
+
+	if ((flags & EROFS_GET_BLOCKS_FIEMAP) ||
+	    ((flags & EROFS_GET_BLOCKS_READMORE) &&
+	     (map->m_algorithmformat == Z_EROFS_COMPRESSION_LZMA ||
+	      map->m_algorithmformat == Z_EROFS_COMPRESSION_DEFLATE ||
+	      map->m_algorithmformat == Z_EROFS_COMPRESSION_ZSTD) &&
+	      map->m_llen >= i_blocksize(inode))) {
+		err = z_erofs_get_extent_decompressedlen(&m);
+		if (!err)
+			map->m_flags &= ~EROFS_MAP_PARTIAL_MAPPED;
+	}
+
+unmap_out:
+	erofs_unmap_metabuf(&m.map->buf);
+	return err;
+}
+
+static int z_erofs_map_blocks_ext(struct inode *inode,
+				  struct erofs_map_blocks *map, int flags)
+{
+	struct erofs_inode *vi = EROFS_I(inode);
+	struct super_block *sb = inode->i_sb;
+	bool interlaced = vi->z_advise & Z_EROFS_ADVISE_INTERLACED_PCLUSTER;
+	unsigned int recsz = z_erofs_extent_recsize(vi->z_advise);
+	erofs_off_t pos = round_up(Z_EROFS_MAP_HEADER_END(erofs_iloc(inode) +
+				   vi->inode_isize + vi->xattr_isize), recsz);
+	unsigned int bmask = sb->s_blocksize - 1;
+	bool in_mbox = erofs_inode_in_metabox(inode);
+	erofs_off_t lend = inode->i_size;
+	erofs_off_t l, r, mid, pa, la, lstart;
+	struct z_erofs_extent *ext;
+	unsigned int fmt;
+	bool last;
+
+	map->m_flags = 0;
+	if (recsz <= offsetof(struct z_erofs_extent, pstart_hi)) {
+		if (recsz <= offsetof(struct z_erofs_extent, pstart_lo)) {
+			ext = erofs_read_metabuf(&map->buf, sb, pos, in_mbox);
+			if (IS_ERR(ext))
+				return PTR_ERR(ext);
+			pa = le64_to_cpu(*(__le64 *)ext);
+			pos += sizeof(__le64);
+			lstart = 0;
+		} else {
+			lstart = round_down(map->m_la, 1 << vi->z_lclusterbits);
+			pos += (lstart >> vi->z_lclusterbits) * recsz;
+			pa = EROFS_NULL_ADDR;
+		}
+
+		for (; lstart <= map->m_la; lstart += 1 << vi->z_lclusterbits) {
+			ext = erofs_read_metabuf(&map->buf, sb, pos, in_mbox);
+			if (IS_ERR(ext))
+				return PTR_ERR(ext);
+			map->m_plen = le32_to_cpu(ext->plen);
+			if (pa != EROFS_NULL_ADDR) {
+				map->m_pa = pa;
+				pa += map->m_plen & Z_EROFS_EXTENT_PLEN_MASK;
+			} else {
+				map->m_pa = le32_to_cpu(ext->pstart_lo);
+			}
+			pos += recsz;
+		}
+		last = (lstart >= round_up(lend, 1 << vi->z_lclusterbits));
+		lend = min(lstart, lend);
+		lstart -= 1 << vi->z_lclusterbits;
+	} else {
+		lstart = lend;
+		for (l = 0, r = vi->z_extents; l < r; ) {
+			mid = l + (r - l) / 2;
+			ext = erofs_read_metabuf(&map->buf, sb,
+						 pos + mid * recsz, in_mbox);
+			if (IS_ERR(ext))
+				return PTR_ERR(ext);
+
+			la = le32_to_cpu(ext->lstart_lo);
+			pa = le32_to_cpu(ext->pstart_lo) |
+				(u64)le32_to_cpu(ext->pstart_hi) << 32;
+			if (recsz > offsetof(struct z_erofs_extent, lstart_hi))
+				la |= (u64)le32_to_cpu(ext->lstart_hi) << 32;
+
+			if (la > map->m_la) {
+				r = mid;
+				if (la > lend) {
+					DBG_BUGON(1);
+					return -EFSCORRUPTED;
+				}
+				lend = la;
+			} else {
+				l = mid + 1;
+				if (map->m_la == la)
+					r = min(l + 1, r);
+				lstart = la;
+				map->m_plen = le32_to_cpu(ext->plen);
+				map->m_pa = pa;
+			}
+		}
+		last = (l >= vi->z_extents);
+	}
+
+	if (lstart < lend) {
+		map->m_la = lstart;
+		if (last && (vi->z_advise & Z_EROFS_ADVISE_FRAGMENT_PCLUSTER)) {
+			map->m_flags = EROFS_MAP_FRAGMENT;
+			vi->z_fragmentoff = map->m_plen;
+			if (recsz > offsetof(struct z_erofs_extent, pstart_lo))
+				vi->z_fragmentoff |= map->m_pa << 32;
+		} else if (map->m_plen & Z_EROFS_EXTENT_PLEN_MASK) {
+			map->m_flags |= EROFS_MAP_MAPPED;
+			fmt = map->m_plen >> Z_EROFS_EXTENT_PLEN_FMT_BIT;
+			if (map->m_plen & Z_EROFS_EXTENT_PLEN_PARTIAL)
+				map->m_flags |= EROFS_MAP_PARTIAL_REF;
+			map->m_plen &= Z_EROFS_EXTENT_PLEN_MASK;
+			if (fmt) {
+				map->m_algorithmformat = --fmt;
+				if (fmt >= Z_EROFS_COMPRESSION_MAX) {
+					erofs_err(sb, "unknown algorithm %d @ pos %llu for nid %llu, please upgrade kernel",
+						  fmt, map->m_la, vi->nid);
+					return -EOPNOTSUPP;
+				}
+			} else if (interlaced && !((map->m_pa | map->m_plen) & bmask))
+				map->m_algorithmformat =
+					Z_EROFS_COMPRESSION_INTERLACED;
+			else
+				map->m_algorithmformat =
+					Z_EROFS_COMPRESSION_SHIFTED;
+		}
+	}
+	map->m_llen = lend - map->m_la;
+	return 0;
+}
+
+static int z_erofs_fill_inode(struct inode *inode, struct erofs_map_blocks *map)
+{
+	struct erofs_inode *const vi = EROFS_I(inode);
+	struct super_block *const sb = inode->i_sb;
+	struct z_erofs_map_header *h;
+	erofs_off_t pos;
+	int err = 0, nr;
+
+	if (test_bit(EROFS_I_Z_INITED_BIT, &vi->flags)) {
+		/*
+		 * paired with smp_mb() at the end of the function to ensure
+		 * fields will only be observed after the bit is set.
+		 */
+		smp_mb();
+		return 0;
+	}
+
+	if (wait_on_bit_lock(&vi->flags, EROFS_I_BL_Z_BIT, TASK_KILLABLE))
+		return -ERESTARTSYS;
+
+	if (test_bit(EROFS_I_Z_INITED_BIT, &vi->flags))
+		goto out_unlock;
+
+	pos = ALIGN(erofs_iloc(inode) + vi->inode_isize + vi->xattr_isize, 8);
+	h = erofs_read_metabuf(&map->buf, sb, pos, erofs_inode_in_metabox(inode));
+	if (IS_ERR(h)) {
+		err = PTR_ERR(h);
+		goto out_unlock;
+	}
+
+	/*
+	 * if the highest bit of the 8-byte map header is set, the whole file
+	 * is stored in the packed inode. The rest bits keeps z_fragmentoff.
+	 */
+	if (h->h_clusterbits >> Z_EROFS_FRAGMENT_INODE_BIT) {
+		vi->z_advise = Z_EROFS_ADVISE_FRAGMENT_PCLUSTER;
+		vi->z_fragmentoff = le64_to_cpu(*(__le64 *)h) ^ (1ULL << 63);
+		vi->z_tailextent_headlcn = 0;
+		goto done;
+	}
+	vi->z_advise = le16_to_cpu(h->h_advise);
+	vi->z_lclusterbits = sb->s_blocksize_bits + (h->h_clusterbits & 15);
+	if (vi->datalayout == EROFS_INODE_COMPRESSED_FULL &&
+	    (vi->z_advise & Z_EROFS_ADVISE_EXTENTS)) {
+		vi->z_extents = le32_to_cpu(h->h_extents_lo) |
+			((u64)le16_to_cpu(h->h_extents_hi) << 32);
+		goto done;
+	}
+
+	if (vi->z_advise & Z_EROFS_ADVISE_FRAGMENT_PCLUSTER)
+		vi->z_fragmentoff = le32_to_cpu(h->h_fragmentoff);
+	else if (vi->z_advise & Z_EROFS_ADVISE_INLINE_PCLUSTER)
+		vi->z_idata_size = le16_to_cpu(h->h_idata_size);
+	for (nr = 0; nr < 2; ++nr) {
+		vi->z_algofmt[nr] = (h->h_algorithmtype >> (4 * nr)) & 15;
+		if (vi->z_algofmt[nr] >= Z_EROFS_COMPRESSION_MAX) {
+			erofs_err(sb, "unknown HEAD%u format %u for nid %llu, please upgrade kernel",
+				  nr + 1, vi->z_algofmt[nr], vi->nid);
+			err = -EOPNOTSUPP;
+			goto out_unlock;
+		}
+	}
+
+	if (!erofs_sb_has_big_pcluster(EROFS_SB(sb)) &&
+	    vi->z_advise & (Z_EROFS_ADVISE_BIG_PCLUSTER_1 |
+			    Z_EROFS_ADVISE_BIG_PCLUSTER_2)) {
+		erofs_err(sb, "per-inode big pcluster without sb feature for nid %llu",
+			  vi->nid);
+		err = -EFSCORRUPTED;
+		goto out_unlock;
+	}
+	if (vi->datalayout == EROFS_INODE_COMPRESSED_COMPACT &&
+	    !(vi->z_advise & Z_EROFS_ADVISE_BIG_PCLUSTER_1) ^
+	    !(vi->z_advise & Z_EROFS_ADVISE_BIG_PCLUSTER_2)) {
+		erofs_err(sb, "big pcluster head1/2 of compact indexes should be consistent for nid %llu",
+			  vi->nid);
+		err = -EFSCORRUPTED;
+		goto out_unlock;
+	}
+
+	if (vi->z_idata_size ||
+	    (vi->z_advise & Z_EROFS_ADVISE_FRAGMENT_PCLUSTER)) {
+		struct erofs_map_blocks tm = {
+			.buf = __EROFS_BUF_INITIALIZER
+		};
+
+		err = z_erofs_map_blocks_fo(inode, &tm,
+					    EROFS_GET_BLOCKS_FINDTAIL);
+		erofs_put_metabuf(&tm.buf);
+		if (err < 0)
+			goto out_unlock;
+	}
+done:
+	/* paired with smp_mb() at the beginning of the function */
+	smp_mb();
+	set_bit(EROFS_I_Z_INITED_BIT, &vi->flags);
+out_unlock:
+	clear_and_wake_up_bit(EROFS_I_BL_Z_BIT, &vi->flags);
+	return err;
+}
+
+static int z_erofs_map_sanity_check(struct inode *inode,
+				    struct erofs_map_blocks *map)
+{
+	struct erofs_sb_info *sbi = EROFS_I_SB(inode);
+	u64 pend;
+
+	if (!(map->m_flags & EROFS_MAP_MAPPED))
+		return 0;
+
+	DBG_BUGON(map->m_algorithmformat >= Z_EROFS_COMPRESSION_RUNTIME_MAX);
+	if (map->m_algorithmformat < Z_EROFS_COMPRESSION_MAX) {
+		if (!(sbi->available_compr_algs & BIT(map->m_algorithmformat))) {
+			erofs_err(inode->i_sb, "inconsistent algorithmtype %u for nid %llu",
+				  map->m_algorithmformat, EROFS_I(inode)->nid);
+			return -EFSCORRUPTED;
+		}
+		if (EROFS_MAP_FULL(map->m_flags) && map->m_llen < map->m_plen &&
+		    map->m_la + map->m_llen < inode->i_size) {
+			erofs_err(inode->i_sb, "too much compressed data @ la %llu of nid %llu",
+				  map->m_la, EROFS_I(inode)->nid);
+			return -EFSCORRUPTED;
+		}
+	} else if (map->m_llen > map->m_plen) {
+		erofs_err(inode->i_sb, "not enough plain data on disk @ la %llu of nid %llu",
+			  map->m_la, EROFS_I(inode)->nid);
+		return -EFSCORRUPTED;
+	}
+	if (unlikely(map->m_plen > Z_EROFS_PCLUSTER_MAX_SIZE ||
+		     map->m_llen > Z_EROFS_PCLUSTER_MAX_DSIZE))
+		return -EOPNOTSUPP;
+	/* Filesystems beyond 48-bit physical block addresses are invalid */
+	if (unlikely(check_add_overflow(map->m_pa, map->m_plen, &pend) ||
+		     (pend >> sbi->blkszbits) >= BIT_ULL(48)))
+		return -EFSCORRUPTED;
+	return 0;
+}
+
+int z_erofs_map_blocks_iter(struct inode *inode, struct erofs_map_blocks *map,
+			    int flags)
+{
+	struct erofs_inode *const vi = EROFS_I(inode);
+	int err = 0;
+
+	trace_erofs_map_blocks_enter(inode, map, flags);
+	if (map->m_la >= inode->i_size) {	/* post-EOF unmapped extent */
+		map->m_llen = map->m_la + 1 - inode->i_size;
+		map->m_la = inode->i_size;
+		map->m_flags = 0;
+	} else {
+		err = z_erofs_fill_inode(inode, map);
+		if (!err) {
+			if (vi->datalayout == EROFS_INODE_COMPRESSED_FULL &&
+			    (vi->z_advise & Z_EROFS_ADVISE_EXTENTS))
+				err = z_erofs_map_blocks_ext(inode, map, flags);
+			else
+				err = z_erofs_map_blocks_fo(inode, map, flags);
+		}
+		if (!err)
+			err = z_erofs_map_sanity_check(inode, map);
+		if (err)
+			map->m_llen = 0;
+	}
+	trace_erofs_map_blocks_exit(inode, map, flags, err);
+	return err;
+}
+
+static int z_erofs_iomap_begin_report(struct inode *inode, loff_t offset,
+				loff_t length, unsigned int flags,
+				struct iomap *iomap, struct iomap *srcmap)
+{
+	int ret;
+	struct erofs_map_blocks map = { .m_la = offset };
+
+	ret = z_erofs_map_blocks_iter(inode, &map, EROFS_GET_BLOCKS_FIEMAP);
+	erofs_put_metabuf(&map.buf);
+	if (ret < 0)
+		return ret;
+
+	iomap->bdev = inode->i_sb->s_bdev;
+	iomap->offset = map.m_la;
+	iomap->length = map.m_llen;
+	if (map.m_flags & EROFS_MAP_FRAGMENT) {
+		iomap->type = IOMAP_MAPPED;
+		iomap->addr = IOMAP_NULL_ADDR;
+	} else if (map.m_flags & EROFS_MAP_MAPPED) {
+		iomap->type = IOMAP_MAPPED;
+		iomap->addr = map.m_pa;
+	} else {
+		iomap->type = IOMAP_HOLE;
+		iomap->addr = IOMAP_NULL_ADDR;
+		/*
+		 * No strict rule on how to describe extents for post EOF, yet
+		 * we need to do like below. Otherwise, iomap itself will get
+		 * into an endless loop on post EOF.
+		 *
+		 * Calculate the effective offset by subtracting extent start
+		 * (map.m_la) from the requested offset, and add it to length.
+		 * (NB: offset >= map.m_la always)
+		 */
+		if (iomap->offset >= inode->i_size)
+			iomap->length = length + offset - map.m_la;
+	}
+	iomap->flags = 0;
+	return 0;
+}
+
+static DEFINE_IOMAP_ITER_NEXT(z_erofs_iomap_next_report,
+			      z_erofs_iomap_begin_report);
+
+const struct iomap_ops z_erofs_iomap_report_ops = {
+	.iomap_next = z_erofs_iomap_next_report,
+};
diff --git a/fs/erofs/zutil.c b/fs/erofs/zutil.c
new file mode 100644
index 000000000..a0ce9226b
--- /dev/null
+++ b/fs/erofs/zutil.c
@@ -0,0 +1,316 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2018 HUAWEI, Inc.
+ *             https://www.huawei.com/
+ * Copyright (C) 2024 Alibaba Cloud
+ */
+#include "internal.h"
+
+struct z_erofs_gbuf {
+	spinlock_t lock;
+	void *ptr;
+	struct page **pages;
+	unsigned int nrpages;
+};
+
+static struct z_erofs_gbuf *z_erofs_gbufpool, *z_erofs_rsvbuf;
+static unsigned int z_erofs_gbuf_count, z_erofs_gbuf_nrpages,
+		z_erofs_rsv_nrpages;
+
+module_param_named(global_buffers, z_erofs_gbuf_count, uint, 0444);
+module_param_named(reserved_pages, z_erofs_rsv_nrpages, uint, 0444);
+
+atomic_long_t erofs_global_shrink_cnt;	/* for all mounted instances */
+
+/* protects `erofs_sb_list_lock` and the mounted `erofs_sb_list` */
+static DEFINE_SPINLOCK(erofs_sb_list_lock);
+static LIST_HEAD(erofs_sb_list);
+static unsigned int shrinker_run_no;
+static struct shrinker *erofs_shrinker_info;
+
+static unsigned int z_erofs_gbuf_id(void)
+{
+	return raw_smp_processor_id() % z_erofs_gbuf_count;
+}
+
+void *z_erofs_get_gbuf(unsigned int requiredpages)
+	__acquires(gbuf->lock)
+{
+	struct z_erofs_gbuf *gbuf;
+
+	migrate_disable();
+	gbuf = &z_erofs_gbufpool[z_erofs_gbuf_id()];
+	spin_lock(&gbuf->lock);
+	/* check if the buffer is too small */
+	if (requiredpages > gbuf->nrpages) {
+		spin_unlock(&gbuf->lock);
+		migrate_enable();
+		/* (for sparse checker) pretend gbuf->lock is still taken */
+		__acquire(gbuf->lock);
+		return NULL;
+	}
+	return gbuf->ptr;
+}
+
+void z_erofs_put_gbuf(void *ptr) __releases(gbuf->lock)
+{
+	struct z_erofs_gbuf *gbuf;
+
+	gbuf = &z_erofs_gbufpool[z_erofs_gbuf_id()];
+	DBG_BUGON(gbuf->ptr != ptr);
+	spin_unlock(&gbuf->lock);
+	migrate_enable();
+}
+
+int z_erofs_gbuf_growsize(unsigned int nrpages)
+{
+	static DEFINE_MUTEX(gbuf_resize_mutex);
+	struct page **tmp_pages = NULL;
+	struct z_erofs_gbuf *gbuf;
+	void *ptr, *old_ptr;
+	int last, i, j;
+
+	mutex_lock(&gbuf_resize_mutex);
+	/* avoid shrinking gbufs, since no idea how many fses rely on */
+	if (nrpages <= z_erofs_gbuf_nrpages) {
+		mutex_unlock(&gbuf_resize_mutex);
+		return 0;
+	}
+
+	for (i = 0; i < z_erofs_gbuf_count; ++i) {
+		gbuf = &z_erofs_gbufpool[i];
+		tmp_pages = kzalloc_objs(*tmp_pages, nrpages);
+		if (!tmp_pages)
+			goto out;
+
+		for (j = 0; j < gbuf->nrpages; ++j)
+			tmp_pages[j] = gbuf->pages[j];
+		do {
+			last = j;
+			j = alloc_pages_bulk(GFP_KERNEL, nrpages,
+					     tmp_pages);
+			if (last == j)
+				goto out;
+		} while (j != nrpages);
+
+		ptr = vmap(tmp_pages, nrpages, VM_MAP, PAGE_KERNEL);
+		if (!ptr)
+			goto out;
+
+		spin_lock(&gbuf->lock);
+		kfree(gbuf->pages);
+		gbuf->pages = tmp_pages;
+		old_ptr = gbuf->ptr;
+		gbuf->ptr = ptr;
+		gbuf->nrpages = nrpages;
+		spin_unlock(&gbuf->lock);
+		if (old_ptr)
+			vunmap(old_ptr);
+	}
+	z_erofs_gbuf_nrpages = nrpages;
+out:
+	if (i < z_erofs_gbuf_count && tmp_pages) {
+		for (j = 0; j < nrpages; ++j)
+			if (tmp_pages[j] && (j >= gbuf->nrpages ||
+					     tmp_pages[j] != gbuf->pages[j]))
+				__free_page(tmp_pages[j]);
+		kfree(tmp_pages);
+	}
+	mutex_unlock(&gbuf_resize_mutex);
+	return i < z_erofs_gbuf_count ? -ENOMEM : 0;
+}
+
+int __init z_erofs_gbuf_init(void)
+{
+	unsigned int i, total = num_possible_cpus();
+
+	if (z_erofs_gbuf_count)
+		total = min(z_erofs_gbuf_count, total);
+	z_erofs_gbuf_count = total;
+
+	/* The last (special) global buffer is the reserved buffer */
+	total += !!z_erofs_rsv_nrpages;
+
+	z_erofs_gbufpool = kzalloc_objs(*z_erofs_gbufpool, total);
+	if (!z_erofs_gbufpool)
+		return -ENOMEM;
+
+	if (z_erofs_rsv_nrpages) {
+		z_erofs_rsvbuf = &z_erofs_gbufpool[total - 1];
+		z_erofs_rsvbuf->pages = kzalloc_objs(*z_erofs_rsvbuf->pages,
+						     z_erofs_rsv_nrpages);
+		if (!z_erofs_rsvbuf->pages) {
+			z_erofs_rsvbuf = NULL;
+			z_erofs_rsv_nrpages = 0;
+		}
+	}
+	for (i = 0; i < total; ++i)
+		spin_lock_init(&z_erofs_gbufpool[i].lock);
+	return 0;
+}
+
+void z_erofs_gbuf_exit(void)
+{
+	int i, j;
+
+	for (i = 0; i < z_erofs_gbuf_count + (!!z_erofs_rsvbuf); ++i) {
+		struct z_erofs_gbuf *gbuf = &z_erofs_gbufpool[i];
+
+		if (gbuf->ptr) {
+			vunmap(gbuf->ptr);
+			gbuf->ptr = NULL;
+		}
+
+		if (!gbuf->pages)
+			continue;
+
+		for (j = 0; j < gbuf->nrpages; ++j)
+			if (gbuf->pages[j])
+				put_page(gbuf->pages[j]);
+		kfree(gbuf->pages);
+		gbuf->pages = NULL;
+	}
+	kfree(z_erofs_gbufpool);
+}
+
+struct page *__erofs_allocpage(struct page **pagepool, gfp_t gfp, bool tryrsv)
+{
+	struct page *page = *pagepool;
+
+	if (page) {
+		*pagepool = (struct page *)page_private(page);
+	} else if (tryrsv && z_erofs_rsvbuf && z_erofs_rsvbuf->nrpages) {
+		spin_lock(&z_erofs_rsvbuf->lock);
+		if (z_erofs_rsvbuf->nrpages)
+			page = z_erofs_rsvbuf->pages[--z_erofs_rsvbuf->nrpages];
+		spin_unlock(&z_erofs_rsvbuf->lock);
+	}
+	if (!page)
+		page = alloc_page(gfp);
+	DBG_BUGON(page && page_ref_count(page) != 1);
+	return page;
+}
+
+void erofs_release_pages(struct page **pagepool)
+{
+	while (*pagepool) {
+		struct page *page = *pagepool;
+
+		*pagepool = (struct page *)page_private(page);
+		/* try to fill reserved global pool first */
+		if (z_erofs_rsvbuf && z_erofs_rsvbuf->nrpages <
+				z_erofs_rsv_nrpages) {
+			spin_lock(&z_erofs_rsvbuf->lock);
+			if (z_erofs_rsvbuf->nrpages < z_erofs_rsv_nrpages) {
+				z_erofs_rsvbuf->pages[z_erofs_rsvbuf->nrpages++]
+						= page;
+				spin_unlock(&z_erofs_rsvbuf->lock);
+				continue;
+			}
+			spin_unlock(&z_erofs_rsvbuf->lock);
+		}
+		put_page(page);
+	}
+}
+
+void erofs_shrinker_register(struct super_block *sb)
+{
+	struct erofs_sb_info *sbi = EROFS_SB(sb);
+
+	mutex_init(&sbi->umount_mutex);
+
+	spin_lock(&erofs_sb_list_lock);
+	list_add(&sbi->list, &erofs_sb_list);
+	spin_unlock(&erofs_sb_list_lock);
+}
+
+void erofs_shrinker_unregister(struct super_block *sb)
+{
+	struct erofs_sb_info *const sbi = EROFS_SB(sb);
+
+	mutex_lock(&sbi->umount_mutex);
+	while (!xa_empty(&sbi->managed_pslots)) {
+		z_erofs_shrink_scan(sbi, ~0UL);
+		cond_resched();
+	}
+	spin_lock(&erofs_sb_list_lock);
+	list_del(&sbi->list);
+	spin_unlock(&erofs_sb_list_lock);
+	mutex_unlock(&sbi->umount_mutex);
+}
+
+static unsigned long erofs_shrink_count(struct shrinker *shrink,
+					struct shrink_control *sc)
+{
+	return atomic_long_read(&erofs_global_shrink_cnt) ?: SHRINK_EMPTY;
+}
+
+static unsigned long erofs_shrink_scan(struct shrinker *shrink,
+				       struct shrink_control *sc)
+{
+	struct erofs_sb_info *sbi;
+	struct list_head *p;
+
+	unsigned long nr = sc->nr_to_scan;
+	unsigned int run_no;
+	unsigned long freed = 0;
+
+	spin_lock(&erofs_sb_list_lock);
+	do {
+		run_no = ++shrinker_run_no;
+	} while (run_no == 0);
+
+	/* Iterate over all mounted superblocks and try to shrink them */
+	p = erofs_sb_list.next;
+	while (p != &erofs_sb_list) {
+		sbi = list_entry(p, struct erofs_sb_info, list);
+
+		/*
+		 * We move the ones we do to the end of the list, so we stop
+		 * when we see one we have already done.
+		 */
+		if (sbi->shrinker_run_no == run_no)
+			break;
+
+		if (!mutex_trylock(&sbi->umount_mutex)) {
+			p = p->next;
+			continue;
+		}
+
+		spin_unlock(&erofs_sb_list_lock);
+		sbi->shrinker_run_no = run_no;
+		freed += z_erofs_shrink_scan(sbi, nr - freed);
+		spin_lock(&erofs_sb_list_lock);
+		/* Get the next list element before we move this one */
+		p = p->next;
+
+		/*
+		 * Move this one to the end of the list to provide some
+		 * fairness.
+		 */
+		list_move_tail(&sbi->list, &erofs_sb_list);
+		mutex_unlock(&sbi->umount_mutex);
+
+		if (freed >= nr)
+			break;
+	}
+	spin_unlock(&erofs_sb_list_lock);
+	return freed;
+}
+
+int __init erofs_init_shrinker(void)
+{
+	erofs_shrinker_info = shrinker_alloc(0, "erofs-shrinker");
+	if (!erofs_shrinker_info)
+		return -ENOMEM;
+
+	erofs_shrinker_info->count_objects = erofs_shrink_count;
+	erofs_shrinker_info->scan_objects = erofs_shrink_scan;
+	shrinker_register(erofs_shrinker_info);
+	return 0;
+}
+
+void erofs_exit_shrinker(void)
+{
+	shrinker_free(erofs_shrinker_info);
+}