[NOMERGE] [RFC PATCH 10/12] erofs: introduce xattr & acl support

[Gao Xiang]

This implements xattr and acl functionalities.

Inline and shared xattrs are introduced for flexibility,
In detail, if the same xattr occurs for many times
on numbers of inodes or the value of xattr is so large
that it isn't suitable for the xattr inline,
shared xattr kept in the xattr meta will be used instead.

Signed-off-by: Miao Xie <miaoxie@xxxxxxxxxx>
Signed-off-by: Chao Yu <yuchao0@xxxxxxxxxx>
Signed-off-by: Gao Xiang <gaoxiang25@xxxxxxxxxx>
---
 fs/erofs/Kconfig    |  37 +++
 fs/erofs/Makefile   |   1 +
 fs/erofs/inode.c    |  61 +++++
 fs/erofs/internal.h |  20 ++
 fs/erofs/namei.c    |   7 +
 fs/erofs/super.c    |  29 +++
 fs/erofs/xattr.c    | 678 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 fs/erofs/xattr.h    |  93 +++++++
 8 files changed, 926 insertions(+)
 create mode 100644 fs/erofs/xattr.c
 create mode 100644 fs/erofs/xattr.h

diff --git a/fs/erofs/Kconfig b/fs/erofs/Kconfig
index 1c2f276..c244cf3 100644
--- a/fs/erofs/Kconfig
+++ b/fs/erofs/Kconfig
@@ -24,6 +24,43 @@ config EROFS_FS_DEBUG
 
 	  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
+	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_USE_VM_MAP_RAM
 	bool "EROFS VM_MAP_RAM Support"
 	depends on EROFS_FS
diff --git a/fs/erofs/Makefile b/fs/erofs/Makefile
index 61dfcae..9d7f90a 100644
--- a/fs/erofs/Makefile
+++ b/fs/erofs/Makefile
@@ -4,4 +4,5 @@ EXTRA_CFLAGS += -Wall -DEROFS_VERSION=\"$(EROFS_VERSION)\"
 
 obj-$(CONFIG_EROFS_FS) += erofs.o
 erofs-objs := super.o inode.o data.o namei.o dir.o
+erofs-$(CONFIG_EROFS_FS_XATTR) += xattr.o
 
diff --git a/fs/erofs/inode.c b/fs/erofs/inode.c
index 94f8ce5..7391ef6 100644
--- a/fs/erofs/inode.c
+++ b/fs/erofs/inode.c
@@ -12,6 +12,7 @@
  */
 #include "internal.h"
 #include <linux/slab.h>
+#include "xattr.h"
 
 /* no locking */
 static int read_inode(struct inode *inode, void *data)
@@ -146,15 +147,26 @@ int fill_inode(struct inode *inode, int isdir)
 	if (!err) {
 		/* setup the new inode */
 		if (S_ISREG(inode->i_mode)) {
+#ifdef CONFIG_EROFS_FS_XATTR
+			if (vi->xattr_isize)
+				inode->i_op = &erofs_generic_xattr_iops;
+#endif
 			inode->i_fop = &generic_ro_fops;
 		} else if (S_ISDIR(inode->i_mode)) {
 			inode->i_op =
+#ifdef CONFIG_EROFS_FS_XATTR
+				vi->xattr_isize ? &erofs_dir_xattr_iops :
+#endif
 				&erofs_dir_iops;
 			inode->i_fop = &erofs_dir_fops;
 		} else if (S_ISLNK(inode->i_mode)) {
 			/* by default, page_get_link is used for symlink */
 			inode->i_op =
+#ifdef CONFIG_EROFS_FS_XATTR
+				&erofs_symlink_xattr_iops,
+#else
 				&page_symlink_inode_operations;
+#endif
 			inode_nohighmem(inode);
 		} else {
 			err = -EIO;
@@ -203,6 +215,33 @@ struct inode *erofs_iget(struct super_block *sb,
 	return inode;
 }
 
+#ifdef CONFIG_EROFS_FS_XATTR
+const struct inode_operations erofs_generic_xattr_iops = {
+#if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 9, 0))
+	.getxattr = generic_getxattr,
+	.listxattr = erofs_listxattr,
+#endif
+};
+#endif
+
+#ifdef CONFIG_EROFS_FS_XATTR
+const struct inode_operations erofs_symlink_xattr_iops = {
+#if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 10, 0))
+	.readlink = generic_readlink,
+#endif
+#if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 5, 0))
+	.follow_link = page_follow_link_light,
+	.put_link = page_put_link,
+#else
+	.get_link = page_get_link,
+#endif
+#if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 9, 0))
+	.getxattr = generic_getxattr,
+	.listxattr = erofs_listxattr,
+#endif
+};
+#endif
+
 #if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 2, 0))
 #include <linux/namei.h>
 
@@ -220,3 +259,25 @@ static void *erofs_follow_fast_link(struct dentry *dentry, struct nameidata *nd)
 };
 #endif
 
+#ifdef CONFIG_EROFS_FS_XATTR
+const struct inode_operations erofs_fast_symlink_xattr_iops = {
+#if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 10, 0))
+	.readlink = generic_readlink,
+#endif
+
+#if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 2, 0))
+	.follow_link = erofs_follow_fast_link,
+#else
+#if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 5, 0))
+	.follow_link = simple_follow_link,
+#else
+	.get_link = simple_get_link,
+#endif
+#endif
+#if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 9, 0))
+	.getxattr = generic_getxattr,
+	.listxattr = erofs_listxattr,
+#endif
+};
+#endif
+
diff --git a/fs/erofs/internal.h b/fs/erofs/internal.h
index 2a3cd45..1dd783c 100644
--- a/fs/erofs/internal.h
+++ b/fs/erofs/internal.h
@@ -50,6 +50,9 @@
 struct erofs_sb_info {
 	u32 blocks;
 	u32 meta_blkaddr;
+#ifdef CONFIG_EROFS_FS_XATTR
+	u32 xattr_blkaddr;
+#endif
 
 	/* inode slot unit size in bit shift */
 	unsigned char islotbits;
@@ -72,6 +75,10 @@ struct erofs_sb_info {
 #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 clear_opt(sbi, option)	((sbi)->mount_opt &= ~EROFS_MOUNT_##option)
 #define set_opt(sbi, option)	((sbi)->mount_opt |= EROFS_MOUNT_##option)
 #define test_opt(sbi, option)	((sbi)->mount_opt & EROFS_MOUNT_##option)
@@ -259,14 +266,27 @@ int erofs_namei(struct inode *dir, struct qstr *name,
 extern const struct inode_operations simple_symlink_inode_operations;
 #endif
 
+#ifdef CONFIG_EROFS_FS_XATTR
+extern const struct inode_operations erofs_symlink_xattr_iops;
+extern const struct inode_operations erofs_fast_symlink_xattr_iops;
+#endif
+
 static inline void set_inode_fast_symlink(struct inode *inode)
 {
+#ifdef CONFIG_EROFS_FS_XATTR
+	inode->i_op = &erofs_fast_symlink_xattr_iops;
+#else
 	inode->i_op = &simple_symlink_inode_operations;
+#endif
 }
 
 static inline bool is_inode_fast_symlink(struct inode *inode)
 {
+#ifdef CONFIG_EROFS_FS_XATTR
+	return inode->i_op == &erofs_fast_symlink_xattr_iops;
+#else
 	return inode->i_op == &simple_symlink_inode_operations;
+#endif
 }
 
 static inline void *erofs_vmap(struct page **pages, unsigned int count)
diff --git a/fs/erofs/namei.c b/fs/erofs/namei.c
index 5a2ddb5..ae71584 100644
--- a/fs/erofs/namei.c
+++ b/fs/erofs/namei.c
@@ -11,6 +11,7 @@
  * distribution for more details.
  */
 #include "internal.h"
+#include "xattr.h"
 
 /* based on the value of qn->len is accurate */
 static inline int dirnamecmp(struct qstr *qn,
@@ -239,5 +240,11 @@ struct dentry *erofs_lookup(struct inode *dir,
 
 const struct inode_operations erofs_dir_xattr_iops = {
 	.lookup = erofs_lookup,
+#ifdef CONFIG_EROFS_FS_XATTR
+#if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 9, 0))
+	.getxattr = generic_getxattr,
+	.listxattr = erofs_listxattr,
+#endif
+#endif
 };
 
diff --git a/fs/erofs/super.c b/fs/erofs/super.c
index a2612e2..b41613f 100644
--- a/fs/erofs/super.c
+++ b/fs/erofs/super.c
@@ -14,6 +14,7 @@
 #include <linux/buffer_head.h>
 #include <linux/slab.h>
 #include <linux/statfs.h>
+#include <linux/seq_file.h>
 #include "internal.h"
 
 static struct kmem_cache *erofs_inode_cachep __read_mostly;
@@ -106,6 +107,9 @@ static int superblock_read(struct super_block *sb)
 
 	sbi->blocks = le32_to_cpu(layout->blocks);
 	sbi->meta_blkaddr = le32_to_cpu(layout->meta_blkaddr);
+#ifdef CONFIG_EROFS_FS_XATTR
+	sbi->xattr_blkaddr = le32_to_cpu(layout->xattr_blkaddr);
+#endif
 	sbi->islotbits = ffs(sizeof(struct erofs_inode_v1)) - 1;
 
 	sbi->root_nid = le64_to_cpu(layout->root_nid);
@@ -130,6 +134,13 @@ static int superblock_read(struct super_block *sb)
 
 static void default_options(struct erofs_sb_info *sbi)
 {
+#ifdef CONFIG_EROFS_FS_XATTR
+	set_opt(sbi, XATTR_USER);
+#endif
+
+#ifdef CONFIG_EROFS_FS_POSIX_ACL
+	set_opt(sbi, POSIX_ACL);
+#endif
 }
 
 static int erofs_read_super(struct super_block *sb,
@@ -165,6 +176,10 @@ static int erofs_read_super(struct super_block *sb,
 
 	sb->s_op = &erofs_sops;
 
+#ifdef CONFIG_EROFS_FS_XATTR
+	sb->s_xattr = erofs_xattr_handlers;
+#endif
+
 	/* set erofs default mount options */
 	default_options(sbi);
 
@@ -333,6 +348,20 @@ static int erofs_statfs(struct dentry *dentry, struct kstatfs *buf)
 
 static int erofs_show_options(struct seq_file *seq, struct dentry *root)
 {
+	struct erofs_sb_info *sbi __maybe_unused = EROFS_SB(root->d_sb);
+
+#ifdef CONFIG_EROFS_FS_XATTR
+	if (test_opt(sbi, XATTR_USER))
+		seq_puts(seq, ",user_xattr");
+	else
+		seq_puts(seq, ",nouser_xattr");
+#endif
+#ifdef CONFIG_EROFS_FS_POSIX_ACL
+	if (test_opt(sbi, POSIX_ACL))
+		seq_puts(seq, ",acl");
+	else
+		seq_puts(seq, ",noacl");
+#endif
 	return 0;
 }
 
diff --git a/fs/erofs/xattr.c b/fs/erofs/xattr.c
new file mode 100644
index 0000000..6ee0f0c
--- /dev/null
+++ b/fs/erofs/xattr.c
@@ -0,0 +1,678 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * linux/fs/erofs/xattr.c
+ *
+ * Copyright (C) 2017-2018 HUAWEI, Inc.
+ *             http://www.huawei.com/
+ * Created by Gao Xiang <gaoxiang25@xxxxxxxxxx>
+ *
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file COPYING in the main directory of the Linux
+ * distribution for more details.
+ */
+#include <linux/slab.h>
+#include <linux/security.h>
+#include "xattr.h"
+#include <linux/version.h>
+
+struct xattr_iter {
+	struct super_block *sb;
+	struct page *page;
+	void *kaddr;
+
+	erofs_blk_t blkaddr;
+	unsigned ofs;
+};
+
+static inline void xattr_iter_end(struct xattr_iter *it, bool atomic)
+{
+	/* only init_inode_xattrs use non-atomic once */
+	if (unlikely(!atomic))
+		kunmap(it->page);
+	else
+		kunmap_atomic(it->kaddr);
+	unlock_page(it->page);
+	put_page(it->page);
+}
+
+static void init_inode_xattrs(struct inode *inode)
+{
+	struct xattr_iter it;
+	unsigned i;
+	struct erofs_xattr_ibody_header *ih;
+	struct erofs_sb_info *sbi;
+	struct erofs_vnode *vi;
+	bool atomic_map;
+
+	if (likely(inode_has_inited_xattr(inode)))
+		return;
+
+	vi = EROFS_V(inode);
+	BUG_ON(!vi->xattr_isize);
+
+	sbi = EROFS_I_SB(inode);
+	it.blkaddr = erofs_blknr(iloc(sbi, vi->nid) + vi->inode_isize);
+	it.ofs = erofs_blkoff(iloc(sbi, vi->nid) + vi->inode_isize);
+
+	it.page = erofs_get_inline_page(inode, it.blkaddr);
+	BUG_ON(IS_ERR(it.page));
+
+	/* read in shared xattr array (non-atomic, see kmalloc below) */
+	it.kaddr = kmap(it.page);
+	atomic_map = false;
+
+	ih = (struct erofs_xattr_ibody_header *)(it.kaddr + it.ofs);
+
+	vi->xattr_shared_count = ih->h_shared_count;
+	vi->xattr_shared_xattrs = (unsigned *)kmalloc_array(
+		vi->xattr_shared_count, sizeof(unsigned),
+		GFP_KERNEL | __GFP_NOFAIL);
+
+	/* let's skip ibody header */
+	it.ofs += sizeof(struct erofs_xattr_ibody_header);
+
+	for (i = 0; i < vi->xattr_shared_count; ++i) {
+		if (unlikely(it.ofs >= EROFS_BLKSIZ)) {
+			/* cannot be unaligned */
+			BUG_ON(it.ofs != EROFS_BLKSIZ);
+			xattr_iter_end(&it, atomic_map);
+
+			it.page = erofs_get_meta_page(inode->i_sb,
+				++it.blkaddr, S_ISDIR(inode->i_mode));
+			BUG_ON(IS_ERR(it.page));
+
+			it.kaddr = kmap_atomic(it.page);
+			atomic_map = true;
+			it.ofs = 0;
+		}
+		vi->xattr_shared_xattrs[i] =
+			le32_to_cpu(*(__le32 *)(it.kaddr + it.ofs));
+		it.ofs += sizeof(__le32);
+	}
+	xattr_iter_end(&it, atomic_map);
+
+	inode_set_inited_xattr(inode);
+}
+
+struct xattr_iter_handlers {
+	int (*entry)(struct xattr_iter *, struct erofs_xattr_entry *);
+	int (*name)(struct xattr_iter *, unsigned, char *, unsigned);
+	int (*alloc_buffer)(struct xattr_iter *, unsigned);
+	void (*value)(struct xattr_iter *, unsigned, char *, unsigned);
+};
+
+static void xattr_iter_fixup(struct xattr_iter *it)
+{
+	if (unlikely(it->ofs >= EROFS_BLKSIZ)) {
+		xattr_iter_end(it, true);
+
+		it->blkaddr += erofs_blknr(it->ofs);
+		it->page = erofs_get_meta_page(it->sb, it->blkaddr, false);
+		BUG_ON(IS_ERR(it->page));
+
+		it->kaddr = kmap_atomic(it->page);
+		it->ofs = erofs_blkoff(it->ofs);
+	}
+}
+
+static int inline_xattr_iter_begin(struct xattr_iter *it,
+	struct inode *inode)
+{
+	unsigned xattr_header_sz, inline_xattr_ofs;
+	struct erofs_vnode *vi = EROFS_V(inode);
+	struct erofs_sb_info *sbi = EROFS_SB(inode->i_sb);
+
+	xattr_header_sz = inlinexattr_header_size(inode);
+	if (unlikely(xattr_header_sz >= vi->xattr_isize)) {
+		BUG_ON(xattr_header_sz > vi->xattr_isize);
+		return -ENOATTR;
+	}
+
+	inline_xattr_ofs = vi->inode_isize + xattr_header_sz;
+
+	it->blkaddr = erofs_blknr(iloc(sbi, vi->nid) + inline_xattr_ofs);
+	it->ofs = erofs_blkoff(iloc(sbi, vi->nid) + inline_xattr_ofs);
+
+	it->page = erofs_get_inline_page(inode, it->blkaddr);
+	BUG_ON(IS_ERR(it->page));
+	it->kaddr = kmap_atomic(it->page);
+
+	return vi->xattr_isize - xattr_header_sz;
+}
+
+static int xattr_foreach(struct xattr_iter *it,
+	struct xattr_iter_handlers *op, unsigned *tlimit)
+{
+	struct erofs_xattr_entry entry;
+	unsigned value_sz, processed, slice;
+	int err;
+
+	/* 0. fixup blkaddr, ofs, ipage */
+	xattr_iter_fixup(it);
+
+	/*
+	 * 1. read xattr entry to the memory,
+	 *    since we do EROFS_XATTR_ALIGN
+	 *    therefore entry should be in the page
+	 */
+	entry = *(struct erofs_xattr_entry *)(it->kaddr + it->ofs);
+	if (tlimit != NULL) {
+		unsigned entry_sz = EROFS_XATTR_ENTRY_SIZE(&entry);
+
+		BUG_ON(*tlimit < entry_sz);
+		*tlimit -= entry_sz;
+	}
+
+	it->ofs += sizeof(struct erofs_xattr_entry);
+	value_sz = le16_to_cpu(entry.e_value_size);
+
+	/* handle entry */
+	err = op->entry(it, &entry);
+	if (err) {
+		it->ofs += entry.e_name_len + value_sz;
+		goto out;
+	}
+
+	/* 2. handle xattr name (ofs will finally be at the end of name) */
+	processed = 0;
+
+	while (processed < entry.e_name_len) {
+		if (it->ofs >= EROFS_BLKSIZ) {
+			BUG_ON(it->ofs > EROFS_BLKSIZ);
+
+			xattr_iter_fixup(it);
+			it->ofs = 0;
+		}
+
+		slice = min_t(unsigned, PAGE_SIZE - it->ofs,
+			entry.e_name_len - processed);
+
+		/* handle name */
+		err = op->name(it, processed, it->kaddr + it->ofs, slice);
+		if (err) {
+			it->ofs += entry.e_name_len - processed + value_sz;
+			goto out;
+		}
+
+		it->ofs += slice;
+		processed += slice;
+	}
+
+	/* 3. handle xattr value */
+	processed = 0;
+
+	if (op->alloc_buffer != NULL) {
+		err = op->alloc_buffer(it, value_sz);
+		if (err) {
+			it->ofs += value_sz;
+			goto out;
+		}
+	}
+
+	while (processed < value_sz) {
+		if (it->ofs >= EROFS_BLKSIZ) {
+			BUG_ON(it->ofs > EROFS_BLKSIZ);
+			xattr_iter_fixup(it);
+			it->ofs = 0;
+		}
+
+		slice = min_t(unsigned, PAGE_SIZE - it->ofs,
+			value_sz - processed);
+		op->value(it, processed, it->kaddr + it->ofs, slice);
+		it->ofs += slice;
+		processed += slice;
+	}
+
+out:
+	/* we assume that ofs is aligned with 4 bytes */
+	it->ofs = EROFS_XATTR_ALIGN(it->ofs);
+	return err;
+}
+
+struct getxattr_iter {
+	struct xattr_iter it;
+
+	char *buffer;
+	int buffer_size, index;
+	struct qstr name;
+};
+
+static int xattr_entrymatch(struct xattr_iter *_it,
+	struct erofs_xattr_entry *entry)
+{
+	struct getxattr_iter *it = container_of(_it, struct getxattr_iter, it);
+
+	return (it->index != entry->e_name_index ||
+		it->name.len != entry->e_name_len) ? -ENOATTR : 0;
+}
+
+static int xattr_namematch(struct xattr_iter *_it,
+	unsigned processed, char *buf, unsigned len)
+{
+	struct getxattr_iter *it = container_of(_it, struct getxattr_iter, it);
+
+	return memcmp(buf, it->name.name + processed, len) ? -ENOATTR : 0;
+}
+
+static int xattr_checkbuffer(struct xattr_iter *_it,
+	unsigned value_sz)
+{
+	struct getxattr_iter *it = container_of(_it, struct getxattr_iter, it);
+	int err = it->buffer_size < value_sz ? -ERANGE : 0;
+
+	it->buffer_size = value_sz;
+	return it->buffer == NULL ? 1 : err;
+}
+
+static void xattr_copyvalue(struct xattr_iter *_it,
+	unsigned processed, char *buf, unsigned len)
+{
+	struct getxattr_iter *it = container_of(_it, struct getxattr_iter, it);
+
+	memcpy(it->buffer + processed, buf, len);
+}
+
+struct xattr_iter_handlers find_xattr_handlers = {
+	.entry = xattr_entrymatch,
+	.name = xattr_namematch,
+	.alloc_buffer = xattr_checkbuffer,
+	.value = xattr_copyvalue
+};
+
+static int inline_getxattr(struct inode *inode, struct getxattr_iter *it)
+{
+	int ret;
+	unsigned remaining;
+
+	ret = inline_xattr_iter_begin(&it->it, inode);
+	if (ret < 0)
+		return ret;
+
+	remaining = ret;
+	while (remaining) {
+		if ((ret = xattr_foreach(&it->it,
+			&find_xattr_handlers, &remaining)) >= 0)
+			break;
+	}
+	xattr_iter_end(&it->it, true);
+
+	return ret < 0 ? ret : it->buffer_size;
+}
+
+static int shared_getxattr(struct inode *inode, struct getxattr_iter *it)
+{
+	struct erofs_vnode *vi = EROFS_V(inode);
+	struct erofs_sb_info *sbi = EROFS_SB(inode->i_sb);
+	unsigned i;
+	int ret = -ENOATTR;
+
+	for (i = 0; i < vi->xattr_shared_count; ++i) {
+		erofs_blk_t blkaddr =
+			xattrblock_addr(sbi, vi->xattr_shared_xattrs[i]);
+
+		it->it.ofs = xattrblock_offset(sbi, vi->xattr_shared_xattrs[i]);
+
+		if (!i || blkaddr != it->it.blkaddr) {
+			if (i)
+				xattr_iter_end(&it->it, true);
+
+			it->it.page = erofs_get_meta_page(inode->i_sb,
+				blkaddr, false);
+			BUG_ON(IS_ERR(it->it.page));
+			it->it.kaddr = kmap_atomic(it->it.page);
+			it->it.blkaddr = blkaddr;
+		}
+
+		if ((ret = xattr_foreach(&it->it,
+			&find_xattr_handlers, NULL)) >= 0)
+			break;
+	}
+	if (vi->xattr_shared_count)
+		xattr_iter_end(&it->it, true);
+
+	return ret < 0 ? ret : it->buffer_size;
+}
+
+#if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 5, 0))
+static int erofs_xattr_get_prefix(struct erofs_sb_info *sbi,
+	int type, const char **prefix)
+{
+	switch (type) {
+	case EROFS_XATTR_INDEX_USER:
+		if (!test_opt(sbi, XATTR_USER))
+			return -EOPNOTSUPP;
+		*prefix = XATTR_USER_PREFIX;
+		return XATTR_USER_PREFIX_LEN;
+
+	case EROFS_XATTR_INDEX_TRUSTED:
+		if (!capable(CAP_SYS_ADMIN))
+			return -EPERM;
+		*prefix = XATTR_TRUSTED_PREFIX;
+		return XATTR_TRUSTED_PREFIX_LEN;
+
+	case EROFS_XATTR_INDEX_SECURITY:
+		*prefix = XATTR_SECURITY_PREFIX;
+		return XATTR_SECURITY_PREFIX_LEN;
+	}
+	return -EINVAL;
+}
+
+#if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 4, 0))
+static size_t erofs_xattr_generic_list(struct dentry *dentry, char *list,
+	size_t list_size, const char *name, size_t name_len, int type)
+#else
+static size_t erofs_xattr_generic_list(const struct xattr_handler *handler,
+	struct dentry *dentry, char *list, size_t list_size,
+	const char *name, size_t name_len)
+#endif
+{
+	struct erofs_sb_info *sbi = EROFS_SB(dentry->d_sb);
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4, 4, 0))
+	int type = handler->flags;
+#endif
+	int total_len, prefix_len;
+	const char *prefix;
+
+	prefix_len = erofs_xattr_get_prefix(sbi, type, &prefix);
+	if (prefix_len < 0)
+		return prefix_len;
+
+	total_len = prefix_len + name_len + 1;
+	if (list && total_len <= list_size) {
+		memcpy(list, prefix, prefix_len);
+		memcpy(list + prefix_len, name, name_len);
+		list[prefix_len + name_len] = '\0';
+	}
+	return total_len;
+}
+
+#else
+static bool erofs_xattr_user_list(struct dentry *dentry)
+{
+	return test_opt(EROFS_SB(dentry->d_sb), XATTR_USER);
+}
+
+static bool erofs_xattr_trusted_list(struct dentry *dentry)
+{
+	return capable(CAP_SYS_ADMIN);
+}
+#endif
+
+int erofs_getxattr(struct inode *inode, int index,
+	const char *name,
+	void *buffer, size_t buffer_size)
+{
+	int ret;
+	struct getxattr_iter it;
+
+	if (unlikely(name == NULL))
+		return -EINVAL;
+
+	init_inode_xattrs(inode);
+
+	it.index = index;
+
+	it.name.len = strlen(name);
+	if (it.name.len > EROFS_NAME_LEN)
+		return -ERANGE;
+	it.name.name = name;
+
+	it.buffer = buffer;
+	it.buffer_size = buffer_size;
+
+	it.it.sb = inode->i_sb;
+	ret = inline_getxattr(inode, &it);
+	if (ret == -ENOATTR)
+		ret = shared_getxattr(inode, &it);
+	return ret;
+}
+
+#if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 4, 0))
+static int erofs_xattr_generic_get(struct dentry *dentry,
+	const char *name,
+	void *buffer, size_t size, int type)
+{
+	struct inode *inode = d_inode(dentry);
+#else
+#if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 7, 0))
+static int erofs_xattr_generic_get(const struct xattr_handler *handler,
+	struct dentry *dentry, const char *name, void *buffer,
+	size_t size)
+{
+	struct inode *inode = d_inode(dentry);
+#else
+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)
+{
+#endif
+#endif
+	struct erofs_sb_info *sbi = EROFS_I_SB(inode);
+
+#if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 4, 0))
+	switch (type) {
+#else
+	switch (handler->flags) {
+#endif
+
+	case EROFS_XATTR_INDEX_USER:
+		if (!test_opt(sbi, XATTR_USER))
+			return -EOPNOTSUPP;
+		break;
+	case EROFS_XATTR_INDEX_TRUSTED:
+		if (!capable(CAP_SYS_ADMIN))
+			return -EPERM;
+		break;
+	case EROFS_XATTR_INDEX_SECURITY:
+		break;
+	default:
+		return -EINVAL;
+	}
+
+#if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 4, 0))
+	if (name[0] == '\0')
+		return -EINVAL;
+#endif
+
+#if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 4, 0))
+	return erofs_getxattr(inode, type, name, buffer, size);
+#else
+	return erofs_getxattr(inode, handler->flags, name, buffer, size);
+#endif
+}
+
+const struct xattr_handler erofs_xattr_user_handler = {
+	.prefix	= XATTR_USER_PREFIX,
+	.flags	= EROFS_XATTR_INDEX_USER,
+#if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 5, 0))
+	.list	= erofs_xattr_generic_list,
+#else
+	.list	= erofs_xattr_user_list,
+#endif
+
+	.get	= erofs_xattr_generic_get,
+};
+
+const struct xattr_handler erofs_xattr_trusted_handler = {
+	.prefix	= XATTR_TRUSTED_PREFIX,
+	.flags	= EROFS_XATTR_INDEX_TRUSTED,
+#if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 5, 0))
+	.list	= erofs_xattr_generic_list,
+#else
+	.list	= erofs_xattr_trusted_list,
+#endif
+	.get	= erofs_xattr_generic_get,
+};
+
+#ifdef CONFIG_EROFS_FS_SECURITY
+const struct xattr_handler __maybe_unused erofs_xattr_security_handler = {
+	.prefix	= XATTR_SECURITY_PREFIX,
+	.flags	= EROFS_XATTR_INDEX_SECURITY,
+	.get	= erofs_xattr_generic_get,
+};
+#endif
+
+const struct xattr_handler *erofs_xattr_handlers[] = {
+	&erofs_xattr_user_handler,
+#ifdef CONFIG_EROFS_FS_POSIX_ACL
+	&posix_acl_access_xattr_handler,
+	&posix_acl_default_xattr_handler,
+#endif
+	&erofs_xattr_trusted_handler,
+#ifdef CONFIG_EROFS_FS_SECURITY
+	&erofs_xattr_security_handler,
+#endif
+	NULL,
+};
+
+struct listxattr_iter {
+	struct xattr_iter it;
+
+	struct dentry *dentry;
+	char *buffer;
+	int buffer_size, buffer_ofs;
+};
+
+static int xattr_entrylist(struct xattr_iter *_it,
+	struct erofs_xattr_entry *entry)
+{
+	struct listxattr_iter *it =
+		container_of(_it, struct listxattr_iter, it);
+	unsigned prefix_len;
+	const char *prefix;
+
+#if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 5, 0))
+	struct erofs_sb_info *sbi = EROFS_SB(it->dentry->d_sb);
+	int ret = erofs_xattr_get_prefix(sbi, entry->e_name_index, &prefix);
+
+	if (ret < 0)
+		return 1;
+
+	prefix_len = ret;
+#else
+	const struct xattr_handler *h =
+		erofs_xattr_handler(entry->e_name_index);
+
+	if (h == NULL || (h->list != NULL && !h->list(it->dentry)))
+		return 1;
+
+	prefix = h->name;
+	prefix_len = h->prefix ? strlen(prefix) : 0;
+#endif
+
+	if (it->buffer == NULL) {
+		it->buffer_ofs += prefix_len + entry->e_name_len + 1;
+		return 1;
+	}
+
+	if (it->buffer_ofs + prefix_len
+		+ entry->e_name_len + 1 > it->buffer_size)
+		return -ERANGE;
+
+	memcpy(it->buffer + it->buffer_ofs, prefix, prefix_len);
+	it->buffer_ofs += prefix_len;
+	return 0;
+}
+
+static int xattr_namelist(struct xattr_iter *_it,
+	unsigned processed, char *buf, unsigned len)
+{
+	struct listxattr_iter *it =
+		container_of(_it, struct listxattr_iter, it);
+
+	memcpy(it->buffer + it->buffer_ofs, buf, len);
+	it->buffer_ofs += len;
+	return 0;
+}
+
+static int xattr_skipvalue(struct xattr_iter *_it,
+	unsigned value_sz)
+{
+	struct listxattr_iter *it =
+		container_of(_it, struct listxattr_iter, it);
+
+	it->buffer[it->buffer_ofs++] = '\0';
+	return 1;
+}
+
+struct xattr_iter_handlers list_xattr_handlers = {
+	.entry = xattr_entrylist,
+	.name = xattr_namelist,
+	.alloc_buffer = xattr_skipvalue,
+	.value = NULL
+};
+
+static int inline_listxattr(struct listxattr_iter *it)
+{
+	int ret;
+	unsigned remaining;
+
+	ret = inline_xattr_iter_begin(&it->it, d_inode(it->dentry));
+	if (ret < 0)
+		return ret;
+
+	remaining = ret;
+	while (remaining) {
+		if ((ret = xattr_foreach(&it->it,
+			&list_xattr_handlers, &remaining)) < 0)
+			break;
+	}
+	xattr_iter_end(&it->it, true);
+	return ret < 0 ? ret : it->buffer_ofs;
+}
+
+static int shared_listxattr(struct listxattr_iter *it)
+{
+	struct inode *inode = d_inode(it->dentry);
+	struct erofs_vnode *vi = EROFS_V(inode);
+	struct erofs_sb_info *sbi = EROFS_I_SB(inode);
+	unsigned i;
+	int ret = 0;
+
+	for (i = 0; i < vi->xattr_shared_count; ++i) {
+		erofs_blk_t blkaddr =
+			xattrblock_addr(sbi, vi->xattr_shared_xattrs[i]);
+
+		it->it.ofs = xattrblock_offset(sbi, vi->xattr_shared_xattrs[i]);
+		if (!i || blkaddr != it->it.blkaddr) {
+			if (i)
+				xattr_iter_end(&it->it, true);
+
+			it->it.page = erofs_get_meta_page(inode->i_sb,
+				blkaddr, false);
+			BUG_ON(IS_ERR(it->it.page));
+			it->it.kaddr = kmap_atomic(it->it.page);
+			it->it.blkaddr = blkaddr;
+		}
+
+		if ((ret = xattr_foreach(&it->it,
+			&list_xattr_handlers, NULL)) < 0)
+			break;
+	}
+	if (vi->xattr_shared_count)
+		xattr_iter_end(&it->it, true);
+
+	return ret < 0 ? ret : it->buffer_ofs;
+}
+
+ssize_t erofs_listxattr(struct dentry *dentry,
+	char *buffer, size_t buffer_size)
+{
+	int ret;
+	struct listxattr_iter it;
+
+	init_inode_xattrs(d_inode(dentry));
+
+	it.dentry = dentry;
+	it.buffer = buffer;
+	it.buffer_size = buffer_size;
+	it.buffer_ofs = 0;
+
+	it.it.sb = dentry->d_sb;
+
+	ret = inline_listxattr(&it);
+	if (ret < 0 && ret != -ENOATTR)
+		return ret;
+	return shared_listxattr(&it);
+}
+
diff --git a/fs/erofs/xattr.h b/fs/erofs/xattr.h
new file mode 100644
index 0000000..b93ccdf
--- /dev/null
+++ b/fs/erofs/xattr.h
@@ -0,0 +1,93 @@
+/* SPDX-License-Identifier: GPL-2.0
+ *
+ * linux/fs/erofs/xattr.h
+ *
+ * Copyright (C) 2017-2018 HUAWEI, Inc.
+ *             http://www.huawei.com/
+ * Created by Gao Xiang <gaoxiang25@xxxxxxxxxx>
+ *
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file COPYING in the main directory of the Linux
+ * distribution for more details.
+ */
+#ifndef __EROFS_XATTR_H
+#define __EROFS_XATTR_H
+
+#include "internal.h"
+#include <linux/posix_acl_xattr.h>
+#include <linux/xattr.h>
+
+/* Attribute not found */
+#define ENOATTR         ENODATA
+
+static inline unsigned inlinexattr_header_size(struct inode *inode)
+{
+	return sizeof(struct erofs_xattr_ibody_header)
+		+ sizeof(u32) * EROFS_V(inode)->xattr_shared_count;
+}
+
+static inline erofs_blk_t
+xattrblock_addr(struct erofs_sb_info *sbi, unsigned xattr_id)
+{
+#ifdef CONFIG_EROFS_FS_XATTR
+	return sbi->xattr_blkaddr +
+		xattr_id * sizeof(__u32) / EROFS_BLKSIZ;
+#else
+	return 0;
+#endif
+}
+
+static inline unsigned
+xattrblock_offset(struct erofs_sb_info *sbi, unsigned xattr_id)
+{
+	return (xattr_id * sizeof(__u32)) % EROFS_BLKSIZ;
+}
+
+extern const struct xattr_handler erofs_xattr_user_handler;
+extern const struct xattr_handler erofs_xattr_trusted_handler;
+#ifdef CONFIG_EROFS_FS_SECURITY
+extern const struct xattr_handler erofs_xattr_security_handler;
+#endif
+
+static inline const struct xattr_handler *erofs_xattr_handler(unsigned index)
+{
+static const struct xattr_handler *xattr_handler_map[] = {
+	[EROFS_XATTR_INDEX_USER] = &erofs_xattr_user_handler,
+#ifdef CONFIG_EROFS_FS_POSIX_ACL
+	[EROFS_XATTR_INDEX_POSIX_ACL_ACCESS] = &posix_acl_access_xattr_handler,
+	[EROFS_XATTR_INDEX_POSIX_ACL_DEFAULT] =
+		&posix_acl_default_xattr_handler,
+#endif
+	[EROFS_XATTR_INDEX_TRUSTED] = &erofs_xattr_trusted_handler,
+#ifdef CONFIG_EROFS_FS_SECURITY
+	[EROFS_XATTR_INDEX_SECURITY] = &erofs_xattr_security_handler,
+#endif
+};
+	return index && index < ARRAY_SIZE(xattr_handler_map) ?
+		xattr_handler_map[index] : NULL;
+}
+
+#ifdef CONFIG_EROFS_FS_XATTR
+
+extern const struct inode_operations erofs_generic_xattr_iops;
+extern const struct inode_operations erofs_dir_xattr_iops;
+
+int erofs_getxattr(struct inode *, int, const char *, void *, size_t);
+ssize_t erofs_listxattr(struct dentry *, char *, size_t);
+#else
+static int __maybe_unused erofs_getxattr(struct inode *inode, int index,
+	const char *name,
+	void *buffer, size_t buffer_size)
+{
+	return -ENOTSUPP;
+}
+
+static ssize_t __maybe_unused erofs_listxattr(struct dentry *dentry,
+	char *buffer, size_t buffer_size)
+{
+	return -ENOTSUPP;
+}
+#endif
+
+#endif
+
-- 
1.9.1