linux-erofs.lists.ozlabs.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] erofs-utils: introduce xattr support
@ 2019-07-26 11:50 htyuxe+dhbrei4sq0df8
  2019-08-05 14:54 ` Li Guifu
  0 siblings, 1 reply; 12+ messages in thread
From: htyuxe+dhbrei4sq0df8 @ 2019-07-26 11:50 UTC (permalink / raw)


load xattrs from source files and pack them into target image.
---
 include/erofs/hashtable.h | 502 ++++++++++++++++++++++++++++++++++++++
 include/erofs/internal.h  |   2 +-
 include/erofs/xattr.h     |  22 ++
 lib/Makefile.am           |   3 +-
 lib/inode.c               |  23 ++
 lib/xattr.c               | 319 ++++++++++++++++++++++++
 6 files changed, 869 insertions(+), 2 deletions(-)
 create mode 100644 include/erofs/hashtable.h
 create mode 100644 include/erofs/xattr.h
 create mode 100644 lib/xattr.c

diff --git a/include/erofs/hashtable.h b/include/erofs/hashtable.h
new file mode 100644
index 0000000..349a655
--- /dev/null
+++ b/include/erofs/hashtable.h
@@ -0,0 +1,502 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * include/erofs/hashtable.h
+ *
+ */
+
+#ifndef __EROFS_HASHTABLE_H
+#define __EROFS_HASHTABLE_H
+
+#define BITS_PER_LONG 32
+#ifndef __always_inline
+#define __always_inline inline
+#endif
+
+/* 2^31 + 2^29 - 2^25 + 2^22 - 2^19 - 2^16 + 1 */
+#define GOLDEN_RATIO_PRIME_32 0x9e370001UL
+/*  2^63 + 2^61 - 2^57 + 2^54 - 2^51 - 2^18 + 1 */
+#define GOLDEN_RATIO_PRIME_64 0x9e37fffffffc0001UL
+
+#if BITS_PER_LONG == 32
+#define GOLDEN_RATIO_PRIME GOLDEN_RATIO_PRIME_32
+#define hash_long(val, bits) hash_32(val, bits)
+#elif BITS_PER_LONG == 64
+#define hash_long(val, bits) hash_64(val, bits)
+#define GOLDEN_RATIO_PRIME GOLDEN_RATIO_PRIME_64
+#else
+#error Wordsize not 32 or 64
+#endif
+
+struct hlist_head {
+	struct hlist_node *first;
+};
+
+struct hlist_node {
+	struct hlist_node *next, **pprev;
+};
+
+/*
+ * Architectures might want to move the poison pointer offset
+ * into some well-recognized area such as 0xdead000000000000,
+ * that is also not mappable by user-space exploits:
+ */
+#ifdef CONFIG_ILLEGAL_POINTER_VALUE
+# define POISON_POINTER_DELTA _AC(CONFIG_ILLEGAL_POINTER_VALUE, UL)
+#else
+# define POISON_POINTER_DELTA 0
+#endif
+
+/*
+ * These are non-NULL pointers that will result in page faults
+ * under normal circumstances, used to verify that nobody uses
+ * non-initialized list entries.
+ */
+#define LIST_POISON1 ((void *) 0x00100100 + POISON_POINTER_DELTA)
+#define LIST_POISON2 ((void *) 0x00200200 + POISON_POINTER_DELTA)
+
+#undef offsetof
+#ifdef __compiler_offsetof
+#define offsetof(TYPE, MEMBER) __compiler_offsetof(TYPE, MEMBER)
+#else
+#define offsetof(TYPE, MEMBER) ((size_t) &((TYPE *)0)->MEMBER)
+#endif
+
+/*
+ * Double linked lists with a single pointer list head.
+ * Mostly useful for hash tables where the two pointer list head is
+ * too wasteful.
+ * You lose the ability to access the tail in O(1).
+ */
+
+#define HLIST_HEAD_INIT { .first = NULL }
+#define HLIST_HEAD(name) struct hlist_head name = { .first = NULL }
+#define INIT_HLIST_HEAD(ptr) ((ptr)->first = NULL)
+static inline void INIT_HLIST_NODE(struct hlist_node *h)
+{
+	h->next = NULL;
+	h->pprev = NULL;
+}
+
+static inline int hlist_unhashed(const struct hlist_node *h)
+{
+	return !h->pprev;
+}
+
+static inline int hlist_empty(const struct hlist_head *h)
+{
+	return !h->first;
+}
+
+static inline void __hlist_del(struct hlist_node *n)
+{
+	struct hlist_node *next = n->next;
+	struct hlist_node **pprev = n->pprev;
+
+	*pprev = next;
+	if (next)
+		next->pprev = pprev;
+}
+
+static inline void hlist_del(struct hlist_node *n)
+{
+	__hlist_del(n);
+	n->next = LIST_POISON1;
+	n->pprev = LIST_POISON2;
+}
+
+static inline void hlist_del_init(struct hlist_node *n)
+{
+	if (!hlist_unhashed(n)) {
+		__hlist_del(n);
+		INIT_HLIST_NODE(n);
+	}
+}
+
+static inline void hlist_add_head(struct hlist_node *n, struct hlist_head *h)
+{
+	struct hlist_node *first = h->first;
+
+	n->next = first;
+	if (first)
+		first->pprev = &n->next;
+	h->first = n;
+	n->pprev = &h->first;
+}
+
+/* next must be != NULL */
+static inline void hlist_add_before(struct hlist_node *n,
+					struct hlist_node *next)
+{
+	n->pprev = next->pprev;
+	n->next = next;
+	next->pprev = &n->next;
+	*(n->pprev) = n;
+}
+
+static inline void hlist_add_behind(struct hlist_node *n,
+				    struct hlist_node *prev)
+{
+	n->next = prev->next;
+	prev->next = n;
+	n->pprev = &prev->next;
+
+	if (n->next)
+		n->next->pprev  = &n->next;
+}
+
+/* after that we'll appear to be on some hlist and hlist_del will work */
+static inline void hlist_add_fake(struct hlist_node *n)
+{
+	n->pprev = &n->next;
+}
+
+/*
+ * Move a list from one list head to another. Fixup the pprev
+ * reference of the first entry if it exists.
+ */
+static inline void hlist_move_list(struct hlist_head *old,
+				   struct hlist_head *new)
+{
+	new->first = old->first;
+	if (new->first)
+		new->first->pprev = &new->first;
+	old->first = NULL;
+}
+
+#define hlist_entry(ptr, type, member) container_of(ptr, type, member)
+
+#define hlist_for_each(pos, head) \
+	for (pos = (head)->first; pos; pos = pos->next)
+
+#define hlist_for_each_safe(pos, n, head) \
+	for (pos = (head)->first; pos && ({ n = pos->next; 1; }); \
+	     pos = n)
+
+#define hlist_entry_safe(ptr, type, member) \
+	({ typeof(ptr) ____ptr = (ptr); \
+	   ____ptr ? hlist_entry(____ptr, type, member) : NULL; \
+	})
+
+/**
+ * hlist_for_each_entry	- iterate over list of given type
+ * @pos:the type * to use as a loop cursor.
+ * @head:the head for your list.
+ * @member:the name of the hlist_node within the struct.
+ */
+#define hlist_for_each_entry(pos, head, member)				\
+	for (pos = hlist_entry_safe((head)->first, typeof(*(pos)), member);\
+	     pos;							\
+	     pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member))
+
+/**
+ * hlist_for_each_entry_continue
+ * iterate over a hlist continuing after current point
+ * @pos:the type * to use as a loop cursor.
+ * @member:the name of the hlist_node within the struct.
+ */
+#define hlist_for_each_entry_continue(pos, member)			\
+	for (pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member);\
+	     pos;							\
+	     pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member))
+
+/**
+ * hlist_for_each_entry_from
+ * iterate over a hlist continuing from current point
+ * @pos:	the type * to use as a loop cursor.
+ * @member:	the name of the hlist_node within the struct.
+ */
+#define hlist_for_each_entry_from(pos, member)				\
+	for (; pos;							\
+	     pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member))
+
+/**
+ * hlist_for_each_entry_safe
+ * iterate over list of given type safe against removal of list entry
+ * @pos:the type * to use as a loop cursor.
+ * @n:another &struct hlist_node to use as temporary storage
+ * @head:the head for your list.
+ * @member:the name of the hlist_node within the struct.
+ */
+#define hlist_for_each_entry_safe(pos, n, head, member) 		\
+	for (pos = hlist_entry_safe((head)->first, typeof(*pos), member);\
+		pos && ({ n = pos->member.next; 1; });			\
+		pos = hlist_entry_safe(n, typeof(*pos), member))
+
+static __always_inline u64 hash_64(u64 val, unsigned int bits)
+{
+	u64 hash = val;
+
+#if defined(CONFIG_ARCH_HAS_FAST_MULTIPLIER) && BITS_PER_LONG == 64
+	hash = hash * GOLDEN_RATIO_PRIME_64;
+#else
+	/*  Sigh, gcc can't optimise this alone like it does for 32 bits. */
+	u64 n = hash;
+
+	n <<= 18;
+	hash -= n;
+	n <<= 33;
+	hash -= n;
+	n <<= 3;
+	hash += n;
+	n <<= 3;
+	hash -= n;
+	n <<= 4;
+	hash += n;
+	n <<= 2;
+	hash += n;
+#endif
+
+	/* High bits are more random, so use them. */
+	return hash >> (64 - bits);
+}
+
+static inline u32 hash_32(u32 val, unsigned int bits)
+{
+	/* On some cpus multiply is faster, on others gcc will do shifts */
+	u32 hash = val * GOLDEN_RATIO_PRIME_32;
+
+	/* High bits are more random, so use them. */
+	return hash >> (32 - bits);
+}
+
+/**
+ * ilog2 - log of base 2 of 32-bit or a 64-bit unsigned value
+ * @n - parameter
+ *
+ * constant-capable log of base 2 calculation
+ * - this can be used to initialise global variables from constant data, hence
+ *   the massive ternary operator construction
+ *
+ * selects the appropriately-sized optimised version depending on sizeof(n)
+ */
+#define ilog2(n)				\
+(								\
+	(n) & (1ULL << 63) ? 63 :	\
+	(n) & (1ULL << 62) ? 62 :	\
+	(n) & (1ULL << 61) ? 61 :	\
+	(n) & (1ULL << 60) ? 60 :	\
+	(n) & (1ULL << 59) ? 59 :	\
+	(n) & (1ULL << 58) ? 58 :	\
+	(n) & (1ULL << 57) ? 57 :	\
+	(n) & (1ULL << 56) ? 56 :	\
+	(n) & (1ULL << 55) ? 55 :	\
+	(n) & (1ULL << 54) ? 54 :	\
+	(n) & (1ULL << 53) ? 53 :	\
+	(n) & (1ULL << 52) ? 52 :	\
+	(n) & (1ULL << 51) ? 51 :	\
+	(n) & (1ULL << 50) ? 50 :	\
+	(n) & (1ULL << 49) ? 49 :	\
+	(n) & (1ULL << 48) ? 48 :	\
+	(n) & (1ULL << 47) ? 47 :	\
+	(n) & (1ULL << 46) ? 46 :	\
+	(n) & (1ULL << 45) ? 45 :	\
+	(n) & (1ULL << 44) ? 44 :	\
+	(n) & (1ULL << 43) ? 43 :	\
+	(n) & (1ULL << 42) ? 42 :	\
+	(n) & (1ULL << 41) ? 41 :	\
+	(n) & (1ULL << 40) ? 40 :	\
+	(n) & (1ULL << 39) ? 39 :	\
+	(n) & (1ULL << 38) ? 38 :	\
+	(n) & (1ULL << 37) ? 37 :	\
+	(n) & (1ULL << 36) ? 36 :	\
+	(n) & (1ULL << 35) ? 35 :	\
+	(n) & (1ULL << 34) ? 34 :	\
+	(n) & (1ULL << 33) ? 33 :	\
+	(n) & (1ULL << 32) ? 32 :	\
+	(n) & (1ULL << 31) ? 31 :	\
+	(n) & (1ULL << 30) ? 30 :	\
+	(n) & (1ULL << 29) ? 29 :	\
+	(n) & (1ULL << 28) ? 28 :	\
+	(n) & (1ULL << 27) ? 27 :	\
+	(n) & (1ULL << 26) ? 26 :	\
+	(n) & (1ULL << 25) ? 25 :	\
+	(n) & (1ULL << 24) ? 24 :	\
+	(n) & (1ULL << 23) ? 23 :	\
+	(n) & (1ULL << 22) ? 22 :	\
+	(n) & (1ULL << 21) ? 21 :	\
+	(n) & (1ULL << 20) ? 20 :	\
+	(n) & (1ULL << 19) ? 19 :	\
+	(n) & (1ULL << 18) ? 18 :	\
+	(n) & (1ULL << 17) ? 17 :	\
+	(n) & (1ULL << 16) ? 16 :	\
+	(n) & (1ULL << 15) ? 15 :	\
+	(n) & (1ULL << 14) ? 14 :	\
+	(n) & (1ULL << 13) ? 13 :	\
+	(n) & (1ULL << 12) ? 12 :	\
+	(n) & (1ULL << 11) ? 11 :	\
+	(n) & (1ULL << 10) ? 10 :	\
+	(n) & (1ULL <<  9) ?  9 :	\
+	(n) & (1ULL <<  8) ?  8 :	\
+	(n) & (1ULL <<  7) ?  7 :	\
+	(n) & (1ULL <<  6) ?  6 :	\
+	(n) & (1ULL <<  5) ?  5 :	\
+	(n) & (1ULL <<  4) ?  4 :	\
+	(n) & (1ULL <<  3) ?  3 :	\
+	(n) & (1ULL <<  2) ?  2 :	\
+	(n) & (1ULL <<  1) ?  1 : 0	\
+)
+
+static const uint16_t crc16tab[256] = {
+	0x0000, 0x1021, 0x2042, 0x3063, 0x4084, 0x50a5, 0x60c6, 0x70e7,
+	0x8108, 0x9129, 0xa14a, 0xb16b, 0xc18c, 0xd1ad, 0xe1ce, 0xf1ef,
+	0x1231, 0x0210, 0x3273, 0x2252, 0x52b5, 0x4294, 0x72f7, 0x62d6,
+	0x9339, 0x8318, 0xb37b, 0xa35a, 0xd3bd, 0xc39c, 0xf3ff, 0xe3de,
+	0x2462, 0x3443, 0x0420, 0x1401, 0x64e6, 0x74c7, 0x44a4, 0x5485,
+	0xa56a, 0xb54b, 0x8528, 0x9509, 0xe5ee, 0xf5cf, 0xc5ac, 0xd58d,
+	0x3653, 0x2672, 0x1611, 0x0630, 0x76d7, 0x66f6, 0x5695, 0x46b4,
+	0xb75b, 0xa77a, 0x9719, 0x8738, 0xf7df, 0xe7fe, 0xd79d, 0xc7bc,
+	0x48c4, 0x58e5, 0x6886, 0x78a7, 0x0840, 0x1861, 0x2802, 0x3823,
+	0xc9cc, 0xd9ed, 0xe98e, 0xf9af, 0x8948, 0x9969, 0xa90a, 0xb92b,
+	0x5af5, 0x4ad4, 0x7ab7, 0x6a96, 0x1a71, 0x0a50, 0x3a33, 0x2a12,
+	0xdbfd, 0xcbdc, 0xfbbf, 0xeb9e, 0x9b79, 0x8b58, 0xbb3b, 0xab1a,
+	0x6ca6, 0x7c87, 0x4ce4, 0x5cc5, 0x2c22, 0x3c03, 0x0c60, 0x1c41,
+	0xedae, 0xfd8f, 0xcdec, 0xddcd, 0xad2a, 0xbd0b, 0x8d68, 0x9d49,
+	0x7e97, 0x6eb6, 0x5ed5, 0x4ef4, 0x3e13, 0x2e32, 0x1e51, 0x0e70,
+	0xff9f, 0xefbe, 0xdfdd, 0xcffc, 0xbf1b, 0xaf3a, 0x9f59, 0x8f78,
+	0x9188, 0x81a9, 0xb1ca, 0xa1eb, 0xd10c, 0xc12d, 0xf14e, 0xe16f,
+	0x1080, 0x00a1, 0x30c2, 0x20e3, 0x5004, 0x4025, 0x7046, 0x6067,
+	0x83b9, 0x9398, 0xa3fb, 0xb3da, 0xc33d, 0xd31c, 0xe37f, 0xf35e,
+	0x02b1, 0x1290, 0x22f3, 0x32d2, 0x4235, 0x5214, 0x6277, 0x7256,
+	0xb5ea, 0xa5cb, 0x95a8, 0x8589, 0xf56e, 0xe54f, 0xd52c, 0xc50d,
+	0x34e2, 0x24c3, 0x14a0, 0x0481, 0x7466, 0x6447, 0x5424, 0x4405,
+	0xa7db, 0xb7fa, 0x8799, 0x97b8, 0xe75f, 0xf77e, 0xc71d, 0xd73c,
+	0x26d3, 0x36f2, 0x0691, 0x16b0, 0x6657, 0x7676, 0x4615, 0x5634,
+	0xd94c, 0xc96d, 0xf90e, 0xe92f, 0x99c8, 0x89e9, 0xb98a, 0xa9ab,
+	0x5844, 0x4865, 0x7806, 0x6827, 0x18c0, 0x08e1, 0x3882, 0x28a3,
+	0xcb7d, 0xdb5c, 0xeb3f, 0xfb1e, 0x8bf9, 0x9bd8, 0xabbb, 0xbb9a,
+	0x4a75, 0x5a54, 0x6a37, 0x7a16, 0x0af1, 0x1ad0, 0x2ab3, 0x3a92,
+	0xfd2e, 0xed0f, 0xdd6c, 0xcd4d, 0xbdaa, 0xad8b, 0x9de8, 0x8dc9,
+	0x7c26, 0x6c07, 0x5c64, 0x4c45, 0x3ca2, 0x2c83, 0x1ce0, 0x0cc1,
+	0xef1f, 0xff3e, 0xcf5d, 0xdf7c, 0xaf9b, 0xbfba, 0x8fd9, 0x9ff8,
+	0x6e17, 0x7e36, 0x4e55, 0x5e74, 0x2e93, 0x3eb2, 0x0ed1, 0x1ef0
+};
+
+uint16_t crc16(const char *buf, int len)
+{
+	int counter;
+	uint16_t crc = 0;
+
+	for (counter = 0; counter < len; counter++)
+		crc = (crc<<8) ^ crc16tab[((crc>>8) ^ *buf++)&0x00FF];
+	return crc;
+}
+
+#define DEFINE_HASHTABLE(name, bits)					\
+	struct hlist_head name[1 << (bits)] =				\
+			{ [0 ... ((1 << (bits)) - 1)] = HLIST_HEAD_INIT }
+
+#define DECLARE_HASHTABLE(name, bits)					\
+	struct hlist_head name[1 << (bits)]
+
+#define HASH_SIZE(name) (ARRAY_SIZE(name))
+#define HASH_BITS(name) ilog2(HASH_SIZE(name))
+
+/* Use hash_32 when possible to allow for fast 32bit hashing in 64bit kernels*/
+#define hash_min(val, bits)						\
+	(sizeof(val) <= 4 ? hash_32(val, bits) : hash_long(val, bits))
+
+static inline void __hash_init(struct hlist_head *ht, unsigned int sz)
+{
+	unsigned int i;
+
+	for (i = 0; i < sz; i++)
+		INIT_HLIST_HEAD(&ht[i]);
+}
+
+/**
+ * hash_init - initialize a hash table
+ * @hashtable: hashtable to be initialized
+ *
+ * Calculates the size of the hashtable from the given parameter, otherwise
+ * same as hash_init_size.
+ *
+ * This has to be a macro since HASH_BITS() will not work on pointers since
+ * it calculates the size during preprocessing.
+ */
+#define hash_init(hashtable) __hash_init(hashtable, HASH_SIZE(hashtable))
+
+/**
+ * hash_add - add an object to a hashtable
+ * @hashtable: hashtable to add to
+ * @node: the &struct hlist_node of the object to be added
+ * @key: the key of the object to be added
+ */
+#define hash_add(hashtable, node, key)					\
+	hlist_add_head(node, &hashtable[hash_min(key, HASH_BITS(hashtable))])
+
+/**
+ * hash_hashed - check whether an object is in any hashtable
+ * @node: the &struct hlist_node of the object to be checked
+ */
+static inline bool hash_hashed(struct hlist_node *node)
+{
+	return !hlist_unhashed(node);
+}
+
+static inline bool __hash_empty(struct hlist_head *ht, unsigned int sz)
+{
+	unsigned int i;
+
+	for (i = 0; i < sz; i++)
+		if (!hlist_empty(&ht[i]))
+			return false;
+
+	return true;
+}
+
+/**
+ * hash_empty - check whether a hashtable is empty
+ * @hashtable: hashtable to check
+ *
+ * This has to be a macro since HASH_BITS() will not work on pointers since
+ * it calculates the size during preprocessing.
+ */
+#define hash_empty(hashtable) __hash_empty(hashtable, HASH_SIZE(hashtable))
+
+/**
+ * hash_del - remove an object from a hashtable
+ * @node: &struct hlist_node of the object to remove
+ */
+static inline void hash_del(struct hlist_node *node)
+{
+	hlist_del_init(node);
+}
+
+/**
+ * hash_for_each - iterate over a hashtable
+ * @name: hashtable to iterate
+ * @bkt: integer to use as bucket loop cursor
+ * @obj: the type * to use as a loop cursor for each entry
+ * @member: the name of the hlist_node within the struct
+ */
+#define hash_for_each(name, bkt, obj, member)				\
+	for ((bkt) = 0, obj = NULL; obj == NULL && (bkt) < HASH_SIZE(name);\
+			(bkt)++)\
+		hlist_for_each_entry(obj, &name[bkt], member)
+
+/**
+ * hash_for_each_safe - iterate over a hashtable safe against removal of
+ * hash entry
+ * @name: hashtable to iterate
+ * @bkt: integer to use as bucket loop cursor
+ * @tmp: a &struct used for temporary storage
+ * @obj: the type * to use as a loop cursor for each entry
+ * @member: the name of the hlist_node within the struct
+ */
+#define hash_for_each_safe(name, bkt, tmp, obj, member)			\
+	for ((bkt) = 0, obj = NULL; obj == NULL && (bkt) < HASH_SIZE(name);\
+			(bkt)++)\
+		hlist_for_each_entry_safe(obj, tmp, &name[bkt], member)
+
+/**
+ * hash_for_each_possible - iterate over all possible objects hashing to the
+ * same bucket
+ * @name: hashtable to iterate
+ * @obj: the type * to use as a loop cursor for each entry
+ * @member: the name of the hlist_node within the struct
+ * @key: the key of the objects to iterate over
+ */
+#define hash_for_each_possible(name, obj, member, key)			\
+	hlist_for_each_entry(obj, &name[hash_min(key, HASH_BITS(name))], member)
+
+#endif
diff --git a/include/erofs/internal.h b/include/erofs/internal.h
index b7ce6f8..33a72b5 100644
--- a/include/erofs/internal.h
+++ b/include/erofs/internal.h
@@ -59,7 +59,7 @@ struct erofs_sb_info {
 extern struct erofs_sb_info sbi;
 
 struct erofs_inode {
-	struct list_head i_hash, i_subdirs;
+	struct list_head i_hash, i_subdirs, i_xattrs;
 
 	unsigned int i_count;
 	struct erofs_inode *i_parent;
diff --git a/include/erofs/xattr.h b/include/erofs/xattr.h
new file mode 100644
index 0000000..dff9fd6
--- /dev/null
+++ b/include/erofs/xattr.h
@@ -0,0 +1,22 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * include/erofs/xattr.h
+ *
+ */
+
+#ifndef __EROFS_XATTR_H
+#define __EROFS_XATTR_H
+
+#define XATTR_COUNT(_size)	({\
+	u32 __size = le16_to_cpu(_size); \
+	((__size) == 0) ? 0 : \
+	(_size - sizeof(struct erofs_xattr_ibody_header)) / \
+	sizeof(struct erofs_xattr_entry) + 1; })
+
+
+int cust_xattr(struct list_head *hlist);
+int read_xattr_from_src(const char *path, struct list_head *hlist);
+int xattr_entry_size(struct list_head *hlist);
+char *xattr_data(struct list_head *hlist, int size);
+
+#endif
diff --git a/lib/Makefile.am b/lib/Makefile.am
index dea82f7..cbe3243 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -2,7 +2,8 @@
 # Makefile.am
 
 noinst_LTLIBRARIES = liberofs.la
-liberofs_la_SOURCES = config.c io.c cache.c inode.c compress.c compressor.c
+liberofs_la_SOURCES = config.c io.c cache.c inode.c compress.c compressor.c \
+		      xattr.c
 liberofs_la_CFLAGS = -Wall -Werror -I$(top_srcdir)/include
 if ENABLE_LZ4
 liberofs_la_CFLAGS += ${LZ4_CFLAGS}
diff --git a/lib/inode.c b/lib/inode.c
index 8b38270..615f117 100644
--- a/lib/inode.c
+++ b/lib/inode.c
@@ -18,6 +18,7 @@
 #include "erofs/cache.h"
 #include "erofs/io.h"
 #include "erofs/compress.h"
+#include "erofs/xattr.h"
 
 struct erofs_sb_info sbi;
 
@@ -364,8 +365,10 @@ static bool erofs_bh_flush_write_inode(struct erofs_buffer_head *bh)
 	/* let's support v1 currently */
 	struct erofs_inode_v1 v1 = {0};
 	int ret;
+	uint16_t count = XATTR_COUNT(inode->xattr_isize);
 
 	v1.i_advise = cpu_to_le16(0 | (inode->data_mapping_mode << 1));
+	v1.i_xattr_icount = cpu_to_le16(count);
 	v1.i_mode = cpu_to_le16(inode->i_mode);
 	v1.i_nlink = cpu_to_le16(inode->i_nlink);
 	v1.i_size = cpu_to_le32((u32)inode->i_size);
@@ -399,6 +402,20 @@ static bool erofs_bh_flush_write_inode(struct erofs_buffer_head *bh)
 		return false;
 	off += inode->inode_isize;
 
+	if (inode->xattr_isize) {
+		char *pbuf = xattr_data(&inode->i_xattrs, inode->xattr_isize);
+
+		if (IS_ERR(pbuf))
+			return false;
+
+		ret = dev_write(pbuf, off, inode->xattr_isize);
+		free(pbuf);
+		if (ret)
+			return false;
+
+		off += inode->xattr_isize;
+	}
+
 	if (inode->extent_isize) {
 		/* write compression metadata */
 		off = Z_EROFS_VLE_EXTENT_ALIGN(off);
@@ -452,6 +469,7 @@ int erofs_prepare_inode_buffer(struct erofs_inode *inode)
 
 	DBG_BUGON(inode->bh || inode->bh_inline);
 
+	inode->xattr_isize = xattr_entry_size(&inode->i_xattrs);
 	inodesize = inode->inode_isize + inode->xattr_isize +
 		    inode->extent_isize;
 
@@ -612,6 +630,7 @@ struct erofs_inode *erofs_new_inode(void)
 	inode->i_count = 1;
 
 	init_list_head(&inode->i_subdirs);
+	init_list_head(&inode->i_xattrs);
 	inode->xattr_isize = 0;
 	inode->extent_isize = 0;
 
@@ -699,6 +718,10 @@ struct erofs_inode *erofs_mkfs_build_tree(struct erofs_inode *dir)
 	struct dirent *dp;
 	struct erofs_dentry *d;
 
+	ret = read_xattr_from_src(dir->i_srcpath, &dir->i_xattrs);
+	if (ret)
+		return ERR_PTR(ret);
+
 	if (!S_ISDIR(dir->i_mode)) {
 		if (S_ISLNK(dir->i_mode)) {
 			char *const symlink = malloc(dir->i_size);
diff --git a/lib/xattr.c b/lib/xattr.c
new file mode 100644
index 0000000..6278abc
--- /dev/null
+++ b/lib/xattr.c
@@ -0,0 +1,319 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * lib/xattr.c
+ *
+ */
+
+#include <stdlib.h>
+#include <sys/types.h>
+#include <sys/xattr.h>
+#include <linux/xattr.h>
+#include <errno.h>
+#include <string.h>
+
+#include "erofs/defs.h"
+#include "erofs/print.h"
+#include "erofs/list.h"
+#include "erofs/internal.h"
+#include "erofs/hashtable.h"
+#include "err.h"
+#define EROFS_XATTR_HASH_TABLE_BITS 16
+
+struct xattr_item {
+	const char *buf;
+	unsigned int keylen;
+	unsigned int vallen;
+	unsigned int count;
+	u8 index;
+	struct hlist_node node;
+};
+
+struct xattr_list {
+	struct xattr_item *item;
+	struct list_head list;
+};
+
+DECLARE_HASHTABLE(my_hash_table, EROFS_XATTR_HASH_TABLE_BITS);
+
+struct xattr_prefix {
+	const char *prefix;
+	uint16_t prefix_len;
+	u8 index;
+} prefix[] = {
+	{
+		XATTR_USER_PREFIX,
+		XATTR_USER_PREFIX_LEN,
+		EROFS_XATTR_INDEX_USER
+	},
+	{
+		XATTR_SECURITY_PREFIX,
+		XATTR_SYSTEM_PREFIX_LEN,
+		EROFS_XATTR_INDEX_SECURITY
+	},
+	{
+		XATTR_TRUSTED_PREFIX,
+		XATTR_TRUSTED_PREFIX_LEN,
+		EROFS_XATTR_INDEX_TRUSTED
+	},
+	{NULL, 0, 0},
+};
+
+static inline void hxattr_add(struct hlist_node *node, uint16_t key)
+{
+	hash_add(my_hash_table, node, key);
+}
+
+static inline void hxattr_del(struct hlist_node *node)
+{
+	hash_del(node);
+}
+
+static struct xattr_item *hxattr_match(const char *buf, int len, u8 index)
+{
+	struct xattr_item *item;
+	uint16_t mkey = crc16(buf, len);
+
+	hash_for_each_possible(my_hash_table, item, node, mkey) {
+		if (index == item->index &&
+		    len == (item->keylen + item->vallen) &&
+		    !memcmp(buf, item->buf, len)) {
+			return item;
+		}
+	}
+
+	return ERR_PTR(-ENOENT);
+}
+
+static bool match_index(const char *key, u8 *index, uint16_t *len)
+{
+	struct xattr_prefix *p = prefix;
+
+	while (p->prefix) {
+		if (strncmp(p->prefix, key, p->prefix_len)) {
+			*len = p->prefix_len;
+			*index = p->index;
+			return true;
+		}
+		p++;
+	}
+
+	return false;
+}
+
+static struct xattr_item *new_xattr(const char *buf, u8 index,
+				    int keylen, int vallen)
+{
+	struct xattr_item *item = malloc(sizeof(*item));
+
+	if (!item)
+		return ERR_PTR(-ENOMEM);
+
+	memset(item, 0, sizeof(*item));
+	INIT_HLIST_NODE(&item->node);
+	item->buf = buf;
+	item->keylen = keylen;
+	item->vallen = vallen;
+	item->count = 1;
+	item->index = index;
+	if (!item->index)
+		return ERR_PTR(-EPERM);
+
+	return item;
+}
+
+static int xattr_add(struct list_head *hlist, struct xattr_item *item)
+{
+	struct xattr_list *mlist = malloc(sizeof(*mlist));
+
+	if (!mlist)
+		return -ENOMEM;
+
+	init_list_head(&mlist->list);
+	mlist->item = item;
+	list_add(&mlist->list, hlist);
+	return 0;
+}
+
+static struct xattr_item *list_xattr_value(const char *path, const char *key)
+{
+	ssize_t keylen, vallen;
+	char *kxattr;
+	struct xattr_item *item;
+	u8 index;
+	uint16_t prelen, suflen;
+
+	/* Output attribute key.*/
+	erofs_info("path:%s key: [%s] ", path, key);
+
+	keylen = strlen(key);
+	if (!match_index(key, &index, &prelen))
+		return ERR_PTR(-ENODATA);
+
+	BUG_ON(keylen < prelen);
+	/* Determine length of the value.*/
+	vallen = lgetxattr(path, key, NULL, 0);
+	if (vallen == -1)
+		return ERR_PTR(-errno);
+
+	/*
+	 * Allocate value buffer.
+	 * One extra byte is needed to append 0x00.
+	 */
+	suflen = keylen - prelen;
+	kxattr = malloc(suflen + vallen + 1);
+	if (!kxattr)
+		return ERR_PTR(-ENOMEM);
+
+	if (vallen == 0)
+		goto value_0;
+
+	/* Copy value to buffer.*/
+	vallen = lgetxattr(path, key, kxattr + suflen, vallen);
+	if (vallen == -1) {
+		free(kxattr);
+		return ERR_PTR(-errno);
+	}
+
+value_0:
+	memcpy(kxattr, key + prelen, suflen);
+	/* Output attribute value.*/
+	kxattr[suflen + vallen] = '\0';
+	erofs_info("value: [%s]", kxattr + suflen);
+
+	/* kxattr is used at xattr_add(), neednt free if SUCCESS */
+	item = hxattr_match(kxattr, suflen + vallen, index);
+	if (!IS_ERR(item)) {
+		item->count++;
+		free(kxattr);
+		return item;
+	}
+
+	item = new_xattr(kxattr, index, suflen, vallen);
+	if (IS_ERR(item))
+		free(kxattr);
+
+	return item;
+}
+
+int read_xattr_from_src(const char *path, struct list_head *hlist)
+{
+	int ret = 0;
+	char *kbuf, *key;
+	ssize_t buflen = llistxattr(path, NULL, 0);
+
+	if (buflen == -1)
+		return -errno;
+	else if (buflen == 0)
+		return 0;
+
+	/* Allocate the buffer.*/
+	kbuf = malloc(buflen);
+	if (!kbuf)
+		return -errno;
+
+	/* Copy the list of attribute keys to the buffer.*/
+	buflen = llistxattr(path, kbuf, buflen);
+	if (buflen == -1) {
+		ret = -errno;
+		goto exit_err;
+	}
+
+	/*
+	 * Loop over the list of zero terminated strings with the
+	 * attribute keys. Use the remaining buffer length to determine
+	 * the end of the list.
+	 */
+	key = kbuf;
+	while (buflen > 0) {
+		size_t keylen = strlen(key) + 1;
+		struct xattr_item *item = list_xattr_value(path, key);
+
+		if (!item) {
+			ret = -errno;
+			goto exit_err;
+		}
+
+		if (!hash_hashed(&item->node)) {
+			uint16_t mkey;
+
+			mkey = crc16(item->buf, item->keylen + item->vallen);
+			hxattr_add(&item->node, mkey);
+		}
+
+		if (hlist) {
+			ret = xattr_add(hlist, item);
+			if (ret < 0)
+				goto exit_err;
+		}
+
+		buflen -= keylen;
+		key += keylen;
+	}
+
+exit_err:
+	free(kbuf);
+	return ret;
+
+}
+
+int xattr_entry_size(struct list_head *hlist)
+{
+	int sum = 0;
+	struct xattr_list *lst;
+
+	if (list_empty(hlist))
+		return 0;
+
+	list_for_each_entry(lst, hlist, list) {
+		struct xattr_item *item = lst->item;
+
+		sum += sizeof(struct erofs_xattr_entry);
+		sum += item->keylen + item->vallen;
+		sum = EROFS_XATTR_ALIGN(sum);
+	}
+
+	sum += sizeof(struct erofs_xattr_ibody_header);
+
+	return EROFS_XATTR_ALIGN(sum);
+}
+
+char *xattr_data(struct list_head *hlist, int xattr_size)
+{
+	struct xattr_list *lst;
+	char *buf, *pbuf;
+	unsigned int size = 0;
+	struct erofs_xattr_ibody_header header = {
+		.h_checksum = 0,
+		.h_shared_count = 0,
+	};
+
+	erofs_info("xattr_size=%d", xattr_size);
+	buf = malloc(xattr_size);
+	if (!buf)
+		return ERR_PTR(-ENOMEM);
+
+	memset(buf, 0, xattr_size);
+	pbuf = buf + sizeof(struct erofs_xattr_ibody_header);
+
+	list_for_each_entry(lst, hlist, list) {
+		struct erofs_xattr_entry entry;
+		struct xattr_item *item = lst->item;
+
+		entry.e_name_index = item->index;
+		entry.e_name_len = item->keylen;
+		entry.e_value_size = cpu_to_le16(item->vallen);
+
+		BUG_ON(size > xattr_size);
+		memcpy(pbuf + size, &entry, sizeof(entry));
+
+		size += sizeof(struct erofs_xattr_entry);
+		memcpy(pbuf + size, item->buf, item->keylen + item->vallen);
+		size += item->keylen + item->vallen;
+		size = EROFS_XATTR_ALIGN(size);
+	}
+
+	memcpy(buf, &header, sizeof(header));
+
+	return buf;
+}
+
-- 
2.17.1

^ permalink raw reply related	[flat|nested] 12+ messages in thread

* [PATCH] erofs-utils: introduce xattr support
  2019-07-26 11:50 [PATCH] erofs-utils: introduce xattr support htyuxe+dhbrei4sq0df8
@ 2019-08-05 14:54 ` Li Guifu
  2019-08-05 17:30   ` Gao Xiang
  0 siblings, 1 reply; 12+ messages in thread
From: Li Guifu @ 2019-08-05 14:54 UTC (permalink / raw)


Hi
It seems look good, a great step.
A new develop branch is a googd choice to make this more stable
and the hash function would better use a light weight one like 
full_name_hash in the kernel

? 2019/7/26 19:50, htyuxe+dhbrei4sq0df8 at grr.la ??:
> load xattrs from source files and pack them into target image.
> ---
>   include/erofs/hashtable.h | 502 ++++++++++++++++++++++++++++++++++++++
>   include/erofs/internal.h  |   2 +-
>   include/erofs/xattr.h     |  22 ++
>   lib/Makefile.am           |   3 +-
>   lib/inode.c               |  23 ++
>   lib/xattr.c               | 319 ++++++++++++++++++++++++
>   6 files changed, 869 insertions(+), 2 deletions(-)
>   create mode 100644 include/erofs/hashtable.h
>   create mode 100644 include/erofs/xattr.h
>   create mode 100644 lib/xattr.c
> 
> diff --git a/include/erofs/hashtable.h b/include/erofs/hashtable.h
> new file mode 100644
> index 0000000..349a655
> --- /dev/null
> +++ b/include/erofs/hashtable.h
> @@ -0,0 +1,502 @@
> +/* SPDX-License-Identifier: GPL-2.0+ */
> +/*
> + * include/erofs/hashtable.h
> + *
> + */
> +
> +#ifndef __EROFS_HASHTABLE_H
> +#define __EROFS_HASHTABLE_H
> +
> +#define BITS_PER_LONG 32
> +#ifndef __always_inline
> +#define __always_inline inline
> +#endif
> +
> +/* 2^31 + 2^29 - 2^25 + 2^22 - 2^19 - 2^16 + 1 */
> +#define GOLDEN_RATIO_PRIME_32 0x9e370001UL
> +/*  2^63 + 2^61 - 2^57 + 2^54 - 2^51 - 2^18 + 1 */
> +#define GOLDEN_RATIO_PRIME_64 0x9e37fffffffc0001UL
> +
> +#if BITS_PER_LONG == 32
> +#define GOLDEN_RATIO_PRIME GOLDEN_RATIO_PRIME_32
> +#define hash_long(val, bits) hash_32(val, bits)
> +#elif BITS_PER_LONG == 64
> +#define hash_long(val, bits) hash_64(val, bits)
> +#define GOLDEN_RATIO_PRIME GOLDEN_RATIO_PRIME_64
> +#else
> +#error Wordsize not 32 or 64
> +#endif
> +
> +struct hlist_head {
> +	struct hlist_node *first;
> +};
> +
> +struct hlist_node {
> +	struct hlist_node *next, **pprev;
> +};
> +
> +/*
> + * Architectures might want to move the poison pointer offset
> + * into some well-recognized area such as 0xdead000000000000,
> + * that is also not mappable by user-space exploits:
> + */
> +#ifdef CONFIG_ILLEGAL_POINTER_VALUE
> +# define POISON_POINTER_DELTA _AC(CONFIG_ILLEGAL_POINTER_VALUE, UL)
> +#else
> +# define POISON_POINTER_DELTA 0
> +#endif
> +
> +/*
> + * These are non-NULL pointers that will result in page faults
> + * under normal circumstances, used to verify that nobody uses
> + * non-initialized list entries.
> + */
> +#define LIST_POISON1 ((void *) 0x00100100 + POISON_POINTER_DELTA)
> +#define LIST_POISON2 ((void *) 0x00200200 + POISON_POINTER_DELTA)
> +
> +#undef offsetof
> +#ifdef __compiler_offsetof
> +#define offsetof(TYPE, MEMBER) __compiler_offsetof(TYPE, MEMBER)
> +#else
> +#define offsetof(TYPE, MEMBER) ((size_t) &((TYPE *)0)->MEMBER)
> +#endif
> +
> +/*
> + * Double linked lists with a single pointer list head.
> + * Mostly useful for hash tables where the two pointer list head is
> + * too wasteful.
> + * You lose the ability to access the tail in O(1).
> + */
> +
> +#define HLIST_HEAD_INIT { .first = NULL }
> +#define HLIST_HEAD(name) struct hlist_head name = { .first = NULL }
> +#define INIT_HLIST_HEAD(ptr) ((ptr)->first = NULL)
> +static inline void INIT_HLIST_NODE(struct hlist_node *h)
> +{
> +	h->next = NULL;
> +	h->pprev = NULL;
> +}
> +
> +static inline int hlist_unhashed(const struct hlist_node *h)
> +{
> +	return !h->pprev;
> +}
> +
> +static inline int hlist_empty(const struct hlist_head *h)
> +{
> +	return !h->first;
> +}
> +
> +static inline void __hlist_del(struct hlist_node *n)
> +{
> +	struct hlist_node *next = n->next;
> +	struct hlist_node **pprev = n->pprev;
> +
> +	*pprev = next;
> +	if (next)
> +		next->pprev = pprev;
> +}
> +
> +static inline void hlist_del(struct hlist_node *n)
> +{
> +	__hlist_del(n);
> +	n->next = LIST_POISON1;
> +	n->pprev = LIST_POISON2;
> +}
> +
> +static inline void hlist_del_init(struct hlist_node *n)
> +{
> +	if (!hlist_unhashed(n)) {
> +		__hlist_del(n);
> +		INIT_HLIST_NODE(n);
> +	}
> +}
> +
> +static inline void hlist_add_head(struct hlist_node *n, struct hlist_head *h)
> +{
> +	struct hlist_node *first = h->first;
> +
> +	n->next = first;
> +	if (first)
> +		first->pprev = &n->next;
> +	h->first = n;
> +	n->pprev = &h->first;
> +}
> +
> +/* next must be != NULL */
> +static inline void hlist_add_before(struct hlist_node *n,
> +					struct hlist_node *next)
> +{
> +	n->pprev = next->pprev;
> +	n->next = next;
> +	next->pprev = &n->next;
> +	*(n->pprev) = n;
> +}
> +
> +static inline void hlist_add_behind(struct hlist_node *n,
> +				    struct hlist_node *prev)
> +{
> +	n->next = prev->next;
> +	prev->next = n;
> +	n->pprev = &prev->next;
> +
> +	if (n->next)
> +		n->next->pprev  = &n->next;
> +}
> +
> +/* after that we'll appear to be on some hlist and hlist_del will work */
> +static inline void hlist_add_fake(struct hlist_node *n)
> +{
> +	n->pprev = &n->next;
> +}
> +
> +/*
> + * Move a list from one list head to another. Fixup the pprev
> + * reference of the first entry if it exists.
> + */
> +static inline void hlist_move_list(struct hlist_head *old,
> +				   struct hlist_head *new)
> +{
> +	new->first = old->first;
> +	if (new->first)
> +		new->first->pprev = &new->first;
> +	old->first = NULL;
> +}
> +
> +#define hlist_entry(ptr, type, member) container_of(ptr, type, member)
> +
> +#define hlist_for_each(pos, head) \
> +	for (pos = (head)->first; pos; pos = pos->next)
> +
> +#define hlist_for_each_safe(pos, n, head) \
> +	for (pos = (head)->first; pos && ({ n = pos->next; 1; }); \
> +	     pos = n)
> +
> +#define hlist_entry_safe(ptr, type, member) \
> +	({ typeof(ptr) ____ptr = (ptr); \
> +	   ____ptr ? hlist_entry(____ptr, type, member) : NULL; \
> +	})
> +
> +/**
> + * hlist_for_each_entry	- iterate over list of given type
> + * @pos:the type * to use as a loop cursor.
> + * @head:the head for your list.
> + * @member:the name of the hlist_node within the struct.
> + */
> +#define hlist_for_each_entry(pos, head, member)				\
> +	for (pos = hlist_entry_safe((head)->first, typeof(*(pos)), member);\
> +	     pos;							\
> +	     pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member))
> +
> +/**
> + * hlist_for_each_entry_continue
> + * iterate over a hlist continuing after current point
> + * @pos:the type * to use as a loop cursor.
> + * @member:the name of the hlist_node within the struct.
> + */
> +#define hlist_for_each_entry_continue(pos, member)			\
> +	for (pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member);\
> +	     pos;							\
> +	     pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member))
> +
> +/**
> + * hlist_for_each_entry_from
> + * iterate over a hlist continuing from current point
> + * @pos:	the type * to use as a loop cursor.
> + * @member:	the name of the hlist_node within the struct.
> + */
> +#define hlist_for_each_entry_from(pos, member)				\
> +	for (; pos;							\
> +	     pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member))
> +
> +/**
> + * hlist_for_each_entry_safe
> + * iterate over list of given type safe against removal of list entry
> + * @pos:the type * to use as a loop cursor.
> + * @n:another &struct hlist_node to use as temporary storage
> + * @head:the head for your list.
> + * @member:the name of the hlist_node within the struct.
> + */
> +#define hlist_for_each_entry_safe(pos, n, head, member) 		\
> +	for (pos = hlist_entry_safe((head)->first, typeof(*pos), member);\
> +		pos && ({ n = pos->member.next; 1; });			\
> +		pos = hlist_entry_safe(n, typeof(*pos), member))
> +
> +static __always_inline u64 hash_64(u64 val, unsigned int bits)
> +{
> +	u64 hash = val;
> +
> +#if defined(CONFIG_ARCH_HAS_FAST_MULTIPLIER) && BITS_PER_LONG == 64
> +	hash = hash * GOLDEN_RATIO_PRIME_64;
> +#else
> +	/*  Sigh, gcc can't optimise this alone like it does for 32 bits. */
> +	u64 n = hash;
> +
> +	n <<= 18;
> +	hash -= n;
> +	n <<= 33;
> +	hash -= n;
> +	n <<= 3;
> +	hash += n;
> +	n <<= 3;
> +	hash -= n;
> +	n <<= 4;
> +	hash += n;
> +	n <<= 2;
> +	hash += n;
> +#endif
> +
> +	/* High bits are more random, so use them. */
> +	return hash >> (64 - bits);
> +}
> +
> +static inline u32 hash_32(u32 val, unsigned int bits)
> +{
> +	/* On some cpus multiply is faster, on others gcc will do shifts */
> +	u32 hash = val * GOLDEN_RATIO_PRIME_32;
> +
> +	/* High bits are more random, so use them. */
> +	return hash >> (32 - bits);
> +}
> +
> +/**
> + * ilog2 - log of base 2 of 32-bit or a 64-bit unsigned value
> + * @n - parameter
> + *
> + * constant-capable log of base 2 calculation
> + * - this can be used to initialise global variables from constant data, hence
> + *   the massive ternary operator construction
> + *
> + * selects the appropriately-sized optimised version depending on sizeof(n)
> + */
> +#define ilog2(n)				\
> +(								\
> +	(n) & (1ULL << 63) ? 63 :	\
> +	(n) & (1ULL << 62) ? 62 :	\
> +	(n) & (1ULL << 61) ? 61 :	\
> +	(n) & (1ULL << 60) ? 60 :	\
> +	(n) & (1ULL << 59) ? 59 :	\
> +	(n) & (1ULL << 58) ? 58 :	\
> +	(n) & (1ULL << 57) ? 57 :	\
> +	(n) & (1ULL << 56) ? 56 :	\
> +	(n) & (1ULL << 55) ? 55 :	\
> +	(n) & (1ULL << 54) ? 54 :	\
> +	(n) & (1ULL << 53) ? 53 :	\
> +	(n) & (1ULL << 52) ? 52 :	\
> +	(n) & (1ULL << 51) ? 51 :	\
> +	(n) & (1ULL << 50) ? 50 :	\
> +	(n) & (1ULL << 49) ? 49 :	\
> +	(n) & (1ULL << 48) ? 48 :	\
> +	(n) & (1ULL << 47) ? 47 :	\
> +	(n) & (1ULL << 46) ? 46 :	\
> +	(n) & (1ULL << 45) ? 45 :	\
> +	(n) & (1ULL << 44) ? 44 :	\
> +	(n) & (1ULL << 43) ? 43 :	\
> +	(n) & (1ULL << 42) ? 42 :	\
> +	(n) & (1ULL << 41) ? 41 :	\
> +	(n) & (1ULL << 40) ? 40 :	\
> +	(n) & (1ULL << 39) ? 39 :	\
> +	(n) & (1ULL << 38) ? 38 :	\
> +	(n) & (1ULL << 37) ? 37 :	\
> +	(n) & (1ULL << 36) ? 36 :	\
> +	(n) & (1ULL << 35) ? 35 :	\
> +	(n) & (1ULL << 34) ? 34 :	\
> +	(n) & (1ULL << 33) ? 33 :	\
> +	(n) & (1ULL << 32) ? 32 :	\
> +	(n) & (1ULL << 31) ? 31 :	\
> +	(n) & (1ULL << 30) ? 30 :	\
> +	(n) & (1ULL << 29) ? 29 :	\
> +	(n) & (1ULL << 28) ? 28 :	\
> +	(n) & (1ULL << 27) ? 27 :	\
> +	(n) & (1ULL << 26) ? 26 :	\
> +	(n) & (1ULL << 25) ? 25 :	\
> +	(n) & (1ULL << 24) ? 24 :	\
> +	(n) & (1ULL << 23) ? 23 :	\
> +	(n) & (1ULL << 22) ? 22 :	\
> +	(n) & (1ULL << 21) ? 21 :	\
> +	(n) & (1ULL << 20) ? 20 :	\
> +	(n) & (1ULL << 19) ? 19 :	\
> +	(n) & (1ULL << 18) ? 18 :	\
> +	(n) & (1ULL << 17) ? 17 :	\
> +	(n) & (1ULL << 16) ? 16 :	\
> +	(n) & (1ULL << 15) ? 15 :	\
> +	(n) & (1ULL << 14) ? 14 :	\
> +	(n) & (1ULL << 13) ? 13 :	\
> +	(n) & (1ULL << 12) ? 12 :	\
> +	(n) & (1ULL << 11) ? 11 :	\
> +	(n) & (1ULL << 10) ? 10 :	\
> +	(n) & (1ULL <<  9) ?  9 :	\
> +	(n) & (1ULL <<  8) ?  8 :	\
> +	(n) & (1ULL <<  7) ?  7 :	\
> +	(n) & (1ULL <<  6) ?  6 :	\
> +	(n) & (1ULL <<  5) ?  5 :	\
> +	(n) & (1ULL <<  4) ?  4 :	\
> +	(n) & (1ULL <<  3) ?  3 :	\
> +	(n) & (1ULL <<  2) ?  2 :	\
> +	(n) & (1ULL <<  1) ?  1 : 0	\
> +)
> +
> +static const uint16_t crc16tab[256] = {
> +	0x0000, 0x1021, 0x2042, 0x3063, 0x4084, 0x50a5, 0x60c6, 0x70e7,
> +	0x8108, 0x9129, 0xa14a, 0xb16b, 0xc18c, 0xd1ad, 0xe1ce, 0xf1ef,
> +	0x1231, 0x0210, 0x3273, 0x2252, 0x52b5, 0x4294, 0x72f7, 0x62d6,
> +	0x9339, 0x8318, 0xb37b, 0xa35a, 0xd3bd, 0xc39c, 0xf3ff, 0xe3de,
> +	0x2462, 0x3443, 0x0420, 0x1401, 0x64e6, 0x74c7, 0x44a4, 0x5485,
> +	0xa56a, 0xb54b, 0x8528, 0x9509, 0xe5ee, 0xf5cf, 0xc5ac, 0xd58d,
> +	0x3653, 0x2672, 0x1611, 0x0630, 0x76d7, 0x66f6, 0x5695, 0x46b4,
> +	0xb75b, 0xa77a, 0x9719, 0x8738, 0xf7df, 0xe7fe, 0xd79d, 0xc7bc,
> +	0x48c4, 0x58e5, 0x6886, 0x78a7, 0x0840, 0x1861, 0x2802, 0x3823,
> +	0xc9cc, 0xd9ed, 0xe98e, 0xf9af, 0x8948, 0x9969, 0xa90a, 0xb92b,
> +	0x5af5, 0x4ad4, 0x7ab7, 0x6a96, 0x1a71, 0x0a50, 0x3a33, 0x2a12,
> +	0xdbfd, 0xcbdc, 0xfbbf, 0xeb9e, 0x9b79, 0x8b58, 0xbb3b, 0xab1a,
> +	0x6ca6, 0x7c87, 0x4ce4, 0x5cc5, 0x2c22, 0x3c03, 0x0c60, 0x1c41,
> +	0xedae, 0xfd8f, 0xcdec, 0xddcd, 0xad2a, 0xbd0b, 0x8d68, 0x9d49,
> +	0x7e97, 0x6eb6, 0x5ed5, 0x4ef4, 0x3e13, 0x2e32, 0x1e51, 0x0e70,
> +	0xff9f, 0xefbe, 0xdfdd, 0xcffc, 0xbf1b, 0xaf3a, 0x9f59, 0x8f78,
> +	0x9188, 0x81a9, 0xb1ca, 0xa1eb, 0xd10c, 0xc12d, 0xf14e, 0xe16f,
> +	0x1080, 0x00a1, 0x30c2, 0x20e3, 0x5004, 0x4025, 0x7046, 0x6067,
> +	0x83b9, 0x9398, 0xa3fb, 0xb3da, 0xc33d, 0xd31c, 0xe37f, 0xf35e,
> +	0x02b1, 0x1290, 0x22f3, 0x32d2, 0x4235, 0x5214, 0x6277, 0x7256,
> +	0xb5ea, 0xa5cb, 0x95a8, 0x8589, 0xf56e, 0xe54f, 0xd52c, 0xc50d,
> +	0x34e2, 0x24c3, 0x14a0, 0x0481, 0x7466, 0x6447, 0x5424, 0x4405,
> +	0xa7db, 0xb7fa, 0x8799, 0x97b8, 0xe75f, 0xf77e, 0xc71d, 0xd73c,
> +	0x26d3, 0x36f2, 0x0691, 0x16b0, 0x6657, 0x7676, 0x4615, 0x5634,
> +	0xd94c, 0xc96d, 0xf90e, 0xe92f, 0x99c8, 0x89e9, 0xb98a, 0xa9ab,
> +	0x5844, 0x4865, 0x7806, 0x6827, 0x18c0, 0x08e1, 0x3882, 0x28a3,
> +	0xcb7d, 0xdb5c, 0xeb3f, 0xfb1e, 0x8bf9, 0x9bd8, 0xabbb, 0xbb9a,
> +	0x4a75, 0x5a54, 0x6a37, 0x7a16, 0x0af1, 0x1ad0, 0x2ab3, 0x3a92,
> +	0xfd2e, 0xed0f, 0xdd6c, 0xcd4d, 0xbdaa, 0xad8b, 0x9de8, 0x8dc9,
> +	0x7c26, 0x6c07, 0x5c64, 0x4c45, 0x3ca2, 0x2c83, 0x1ce0, 0x0cc1,
> +	0xef1f, 0xff3e, 0xcf5d, 0xdf7c, 0xaf9b, 0xbfba, 0x8fd9, 0x9ff8,
> +	0x6e17, 0x7e36, 0x4e55, 0x5e74, 0x2e93, 0x3eb2, 0x0ed1, 0x1ef0
> +};
> +
> +uint16_t crc16(const char *buf, int len)
> +{
> +	int counter;
> +	uint16_t crc = 0;
> +
> +	for (counter = 0; counter < len; counter++)
> +		crc = (crc<<8) ^ crc16tab[((crc>>8) ^ *buf++)&0x00FF];
> +	return crc;
> +}
> +
> +#define DEFINE_HASHTABLE(name, bits)					\
> +	struct hlist_head name[1 << (bits)] =				\
> +			{ [0 ... ((1 << (bits)) - 1)] = HLIST_HEAD_INIT }
> +
> +#define DECLARE_HASHTABLE(name, bits)					\
> +	struct hlist_head name[1 << (bits)]
> +
> +#define HASH_SIZE(name) (ARRAY_SIZE(name))
> +#define HASH_BITS(name) ilog2(HASH_SIZE(name))
> +
> +/* Use hash_32 when possible to allow for fast 32bit hashing in 64bit kernels*/
> +#define hash_min(val, bits)						\
> +	(sizeof(val) <= 4 ? hash_32(val, bits) : hash_long(val, bits))
> +
> +static inline void __hash_init(struct hlist_head *ht, unsigned int sz)
> +{
> +	unsigned int i;
> +
> +	for (i = 0; i < sz; i++)
> +		INIT_HLIST_HEAD(&ht[i]);
> +}
> +
> +/**
> + * hash_init - initialize a hash table
> + * @hashtable: hashtable to be initialized
> + *
> + * Calculates the size of the hashtable from the given parameter, otherwise
> + * same as hash_init_size.
> + *
> + * This has to be a macro since HASH_BITS() will not work on pointers since
> + * it calculates the size during preprocessing.
> + */
> +#define hash_init(hashtable) __hash_init(hashtable, HASH_SIZE(hashtable))
> +
> +/**
> + * hash_add - add an object to a hashtable
> + * @hashtable: hashtable to add to
> + * @node: the &struct hlist_node of the object to be added
> + * @key: the key of the object to be added
> + */
> +#define hash_add(hashtable, node, key)					\
> +	hlist_add_head(node, &hashtable[hash_min(key, HASH_BITS(hashtable))])
> +
> +/**
> + * hash_hashed - check whether an object is in any hashtable
> + * @node: the &struct hlist_node of the object to be checked
> + */
> +static inline bool hash_hashed(struct hlist_node *node)
> +{
> +	return !hlist_unhashed(node);
> +}
> +
> +static inline bool __hash_empty(struct hlist_head *ht, unsigned int sz)
> +{
> +	unsigned int i;
> +
> +	for (i = 0; i < sz; i++)
> +		if (!hlist_empty(&ht[i]))
> +			return false;
> +
> +	return true;
> +}
> +
> +/**
> + * hash_empty - check whether a hashtable is empty
> + * @hashtable: hashtable to check
> + *
> + * This has to be a macro since HASH_BITS() will not work on pointers since
> + * it calculates the size during preprocessing.
> + */
> +#define hash_empty(hashtable) __hash_empty(hashtable, HASH_SIZE(hashtable))
> +
> +/**
> + * hash_del - remove an object from a hashtable
> + * @node: &struct hlist_node of the object to remove
> + */
> +static inline void hash_del(struct hlist_node *node)
> +{
> +	hlist_del_init(node);
> +}
> +
> +/**
> + * hash_for_each - iterate over a hashtable
> + * @name: hashtable to iterate
> + * @bkt: integer to use as bucket loop cursor
> + * @obj: the type * to use as a loop cursor for each entry
> + * @member: the name of the hlist_node within the struct
> + */
> +#define hash_for_each(name, bkt, obj, member)				\
> +	for ((bkt) = 0, obj = NULL; obj == NULL && (bkt) < HASH_SIZE(name);\
> +			(bkt)++)\
> +		hlist_for_each_entry(obj, &name[bkt], member)
> +
> +/**
> + * hash_for_each_safe - iterate over a hashtable safe against removal of
> + * hash entry
> + * @name: hashtable to iterate
> + * @bkt: integer to use as bucket loop cursor
> + * @tmp: a &struct used for temporary storage
> + * @obj: the type * to use as a loop cursor for each entry
> + * @member: the name of the hlist_node within the struct
> + */
> +#define hash_for_each_safe(name, bkt, tmp, obj, member)			\
> +	for ((bkt) = 0, obj = NULL; obj == NULL && (bkt) < HASH_SIZE(name);\
> +			(bkt)++)\
> +		hlist_for_each_entry_safe(obj, tmp, &name[bkt], member)
> +
> +/**
> + * hash_for_each_possible - iterate over all possible objects hashing to the
> + * same bucket
> + * @name: hashtable to iterate
> + * @obj: the type * to use as a loop cursor for each entry
> + * @member: the name of the hlist_node within the struct
> + * @key: the key of the objects to iterate over
> + */
> +#define hash_for_each_possible(name, obj, member, key)			\
> +	hlist_for_each_entry(obj, &name[hash_min(key, HASH_BITS(name))], member)
> +
> +#endif
> diff --git a/include/erofs/internal.h b/include/erofs/internal.h
> index b7ce6f8..33a72b5 100644
> --- a/include/erofs/internal.h
> +++ b/include/erofs/internal.h
> @@ -59,7 +59,7 @@ struct erofs_sb_info {
>   extern struct erofs_sb_info sbi;
>   
>   struct erofs_inode {
> -	struct list_head i_hash, i_subdirs;
> +	struct list_head i_hash, i_subdirs, i_xattrs;
>   
>   	unsigned int i_count;
>   	struct erofs_inode *i_parent;
> diff --git a/include/erofs/xattr.h b/include/erofs/xattr.h
> new file mode 100644
> index 0000000..dff9fd6
> --- /dev/null
> +++ b/include/erofs/xattr.h
> @@ -0,0 +1,22 @@
> +// SPDX-License-Identifier: GPL-2.0+
> +/*
> + * include/erofs/xattr.h
> + *
> + */
> +
> +#ifndef __EROFS_XATTR_H
> +#define __EROFS_XATTR_H
> +
> +#define XATTR_COUNT(_size)	({\
> +	u32 __size = le16_to_cpu(_size); \
> +	((__size) == 0) ? 0 : \
> +	(_size - sizeof(struct erofs_xattr_ibody_header)) / \
> +	sizeof(struct erofs_xattr_entry) + 1; })
> +
> +
> +int cust_xattr(struct list_head *hlist);
> +int read_xattr_from_src(const char *path, struct list_head *hlist);
> +int xattr_entry_size(struct list_head *hlist);
> +char *xattr_data(struct list_head *hlist, int size);
> +
> +#endif
> diff --git a/lib/Makefile.am b/lib/Makefile.am
> index dea82f7..cbe3243 100644
> --- a/lib/Makefile.am
> +++ b/lib/Makefile.am
> @@ -2,7 +2,8 @@
>   # Makefile.am
>   
>   noinst_LTLIBRARIES = liberofs.la
> -liberofs_la_SOURCES = config.c io.c cache.c inode.c compress.c compressor.c
> +liberofs_la_SOURCES = config.c io.c cache.c inode.c compress.c compressor.c \
> +		      xattr.c
>   liberofs_la_CFLAGS = -Wall -Werror -I$(top_srcdir)/include
>   if ENABLE_LZ4
>   liberofs_la_CFLAGS += ${LZ4_CFLAGS}
> diff --git a/lib/inode.c b/lib/inode.c
> index 8b38270..615f117 100644
> --- a/lib/inode.c
> +++ b/lib/inode.c
> @@ -18,6 +18,7 @@
>   #include "erofs/cache.h"
>   #include "erofs/io.h"
>   #include "erofs/compress.h"
> +#include "erofs/xattr.h"
>   
>   struct erofs_sb_info sbi;
>   
> @@ -364,8 +365,10 @@ static bool erofs_bh_flush_write_inode(struct erofs_buffer_head *bh)
>   	/* let's support v1 currently */
>   	struct erofs_inode_v1 v1 = {0};
>   	int ret;
> +	uint16_t count = XATTR_COUNT(inode->xattr_isize);
>   
>   	v1.i_advise = cpu_to_le16(0 | (inode->data_mapping_mode << 1));
> +	v1.i_xattr_icount = cpu_to_le16(count);
>   	v1.i_mode = cpu_to_le16(inode->i_mode);
>   	v1.i_nlink = cpu_to_le16(inode->i_nlink);
>   	v1.i_size = cpu_to_le32((u32)inode->i_size);
> @@ -399,6 +402,20 @@ static bool erofs_bh_flush_write_inode(struct erofs_buffer_head *bh)
>   		return false;
>   	off += inode->inode_isize;
>   
> +	if (inode->xattr_isize) {
> +		char *pbuf = xattr_data(&inode->i_xattrs, inode->xattr_isize);
> +
> +		if (IS_ERR(pbuf))
> +			return false;
> +
> +		ret = dev_write(pbuf, off, inode->xattr_isize);
> +		free(pbuf);
> +		if (ret)
> +			return false;
> +
> +		off += inode->xattr_isize;
> +	}
> +
>   	if (inode->extent_isize) {
>   		/* write compression metadata */
>   		off = Z_EROFS_VLE_EXTENT_ALIGN(off);
> @@ -452,6 +469,7 @@ int erofs_prepare_inode_buffer(struct erofs_inode *inode)
>   
>   	DBG_BUGON(inode->bh || inode->bh_inline);
>   
> +	inode->xattr_isize = xattr_entry_size(&inode->i_xattrs);
>   	inodesize = inode->inode_isize + inode->xattr_isize +
>   		    inode->extent_isize;
>   
> @@ -612,6 +630,7 @@ struct erofs_inode *erofs_new_inode(void)
>   	inode->i_count = 1;
>   
>   	init_list_head(&inode->i_subdirs);
> +	init_list_head(&inode->i_xattrs);
>   	inode->xattr_isize = 0;
>   	inode->extent_isize = 0;
>   
> @@ -699,6 +718,10 @@ struct erofs_inode *erofs_mkfs_build_tree(struct erofs_inode *dir)
>   	struct dirent *dp;
>   	struct erofs_dentry *d;
>   
> +	ret = read_xattr_from_src(dir->i_srcpath, &dir->i_xattrs);
> +	if (ret)
> +		return ERR_PTR(ret);
> +
>   	if (!S_ISDIR(dir->i_mode)) {
>   		if (S_ISLNK(dir->i_mode)) {
>   			char *const symlink = malloc(dir->i_size);
> diff --git a/lib/xattr.c b/lib/xattr.c
> new file mode 100644
> index 0000000..6278abc
> --- /dev/null
> +++ b/lib/xattr.c
> @@ -0,0 +1,319 @@
> +// SPDX-License-Identifier: GPL-2.0+
> +/*
> + * lib/xattr.c
> + *
> + */
> +
> +#include <stdlib.h>
> +#include <sys/types.h>
> +#include <sys/xattr.h>
> +#include <linux/xattr.h>
> +#include <errno.h>
> +#include <string.h>
> +
> +#include "erofs/defs.h"
> +#include "erofs/print.h"
> +#include "erofs/list.h"
> +#include "erofs/internal.h"
> +#include "erofs/hashtable.h"
> +#include "err.h"
> +#define EROFS_XATTR_HASH_TABLE_BITS 16
> +
> +struct xattr_item {
> +	const char *buf;
> +	unsigned int keylen;
> +	unsigned int vallen;
> +	unsigned int count;
> +	u8 index;
> +	struct hlist_node node;
> +};
> +
> +struct xattr_list {
> +	struct xattr_item *item;
> +	struct list_head list;
> +};
> +
> +DECLARE_HASHTABLE(my_hash_table, EROFS_XATTR_HASH_TABLE_BITS);
> +
> +struct xattr_prefix {
> +	const char *prefix;
> +	uint16_t prefix_len;
> +	u8 index;
> +} prefix[] = {
> +	{
> +		XATTR_USER_PREFIX,
> +		XATTR_USER_PREFIX_LEN,
> +		EROFS_XATTR_INDEX_USER
> +	},
> +	{
> +		XATTR_SECURITY_PREFIX,
> +		XATTR_SYSTEM_PREFIX_LEN,
> +		EROFS_XATTR_INDEX_SECURITY
> +	},
> +	{
> +		XATTR_TRUSTED_PREFIX,
> +		XATTR_TRUSTED_PREFIX_LEN,
> +		EROFS_XATTR_INDEX_TRUSTED
> +	},
> +	{NULL, 0, 0},
> +};
> +
> +static inline void hxattr_add(struct hlist_node *node, uint16_t key)
> +{
> +	hash_add(my_hash_table, node, key);
> +}
> +
> +static inline void hxattr_del(struct hlist_node *node)
> +{
> +	hash_del(node);
> +}
> +
> +static struct xattr_item *hxattr_match(const char *buf, int len, u8 index)
> +{
> +	struct xattr_item *item;
> +	uint16_t mkey = crc16(buf, len);
> +
> +	hash_for_each_possible(my_hash_table, item, node, mkey) {
> +		if (index == item->index &&
> +		    len == (item->keylen + item->vallen) &&
> +		    !memcmp(buf, item->buf, len)) {
> +			return item;
> +		}
> +	}
> +
> +	return ERR_PTR(-ENOENT);
> +}
> +
> +static bool match_index(const char *key, u8 *index, uint16_t *len)
> +{
> +	struct xattr_prefix *p = prefix;
> +
> +	while (p->prefix) {
> +		if (strncmp(p->prefix, key, p->prefix_len)) {
> +			*len = p->prefix_len;
> +			*index = p->index;
> +			return true;
> +		}
> +		p++;
> +	}
> +
> +	return false;
> +}
> +
> +static struct xattr_item *new_xattr(const char *buf, u8 index,
> +				    int keylen, int vallen)
> +{
> +	struct xattr_item *item = malloc(sizeof(*item));
> +
> +	if (!item)
> +		return ERR_PTR(-ENOMEM);
> +
> +	memset(item, 0, sizeof(*item));
> +	INIT_HLIST_NODE(&item->node);
> +	item->buf = buf;
> +	item->keylen = keylen;
> +	item->vallen = vallen;
> +	item->count = 1;
> +	item->index = index;
> +	if (!item->index)
> +		return ERR_PTR(-EPERM);
> +
> +	return item;
> +}
> +
> +static int xattr_add(struct list_head *hlist, struct xattr_item *item)
> +{
> +	struct xattr_list *mlist = malloc(sizeof(*mlist));
> +
> +	if (!mlist)
> +		return -ENOMEM;
> +
> +	init_list_head(&mlist->list);
> +	mlist->item = item;
> +	list_add(&mlist->list, hlist);
> +	return 0;
> +}
> +
> +static struct xattr_item *list_xattr_value(const char *path, const char *key)
> +{
> +	ssize_t keylen, vallen;
> +	char *kxattr;
> +	struct xattr_item *item;
> +	u8 index;
> +	uint16_t prelen, suflen;
> +
> +	/* Output attribute key.*/
> +	erofs_info("path:%s key: [%s] ", path, key);
> +
> +	keylen = strlen(key);
> +	if (!match_index(key, &index, &prelen))
> +		return ERR_PTR(-ENODATA);
> +
> +	BUG_ON(keylen < prelen);
> +	/* Determine length of the value.*/
> +	vallen = lgetxattr(path, key, NULL, 0);
> +	if (vallen == -1)
> +		return ERR_PTR(-errno);
> +
> +	/*
> +	 * Allocate value buffer.
> +	 * One extra byte is needed to append 0x00.
> +	 */
> +	suflen = keylen - prelen;
> +	kxattr = malloc(suflen + vallen + 1);
> +	if (!kxattr)
> +		return ERR_PTR(-ENOMEM);
> +
> +	if (vallen == 0)
> +		goto value_0;
> +
> +	/* Copy value to buffer.*/
> +	vallen = lgetxattr(path, key, kxattr + suflen, vallen);
> +	if (vallen == -1) {
> +		free(kxattr);
> +		return ERR_PTR(-errno);
> +	}
> +
> +value_0:
> +	memcpy(kxattr, key + prelen, suflen);
> +	/* Output attribute value.*/
> +	kxattr[suflen + vallen] = '\0';
> +	erofs_info("value: [%s]", kxattr + suflen);
> +
> +	/* kxattr is used at xattr_add(), neednt free if SUCCESS */
> +	item = hxattr_match(kxattr, suflen + vallen, index);
> +	if (!IS_ERR(item)) {
> +		item->count++;
> +		free(kxattr);
> +		return item;
> +	}
> +
> +	item = new_xattr(kxattr, index, suflen, vallen);
> +	if (IS_ERR(item))
> +		free(kxattr);
> +
> +	return item;
> +}
> +
> +int read_xattr_from_src(const char *path, struct list_head *hlist)
> +{
> +	int ret = 0;
> +	char *kbuf, *key;
> +	ssize_t buflen = llistxattr(path, NULL, 0);
> +
> +	if (buflen == -1)
> +		return -errno;
> +	else if (buflen == 0)
> +		return 0;
> +
> +	/* Allocate the buffer.*/
> +	kbuf = malloc(buflen);
> +	if (!kbuf)
> +		return -errno;
> +
> +	/* Copy the list of attribute keys to the buffer.*/
> +	buflen = llistxattr(path, kbuf, buflen);
> +	if (buflen == -1) {
> +		ret = -errno;
> +		goto exit_err;
> +	}
> +
> +	/*
> +	 * Loop over the list of zero terminated strings with the
> +	 * attribute keys. Use the remaining buffer length to determine
> +	 * the end of the list.
> +	 */
> +	key = kbuf;
> +	while (buflen > 0) {
> +		size_t keylen = strlen(key) + 1;
> +		struct xattr_item *item = list_xattr_value(path, key);
> +
> +		if (!item) {
> +			ret = -errno;
> +			goto exit_err;
> +		}
> +
> +		if (!hash_hashed(&item->node)) {
> +			uint16_t mkey;
> +
> +			mkey = crc16(item->buf, item->keylen + item->vallen);
> +			hxattr_add(&item->node, mkey);
> +		}
> +
> +		if (hlist) {
> +			ret = xattr_add(hlist, item);
> +			if (ret < 0)
> +				goto exit_err;
> +		}
> +
> +		buflen -= keylen;
> +		key += keylen;
> +	}
> +
> +exit_err:
> +	free(kbuf);
> +	return ret;
> +
> +}
> +
> +int xattr_entry_size(struct list_head *hlist)
> +{
> +	int sum = 0;
> +	struct xattr_list *lst;
> +
> +	if (list_empty(hlist))
> +		return 0;
> +
> +	list_for_each_entry(lst, hlist, list) {
> +		struct xattr_item *item = lst->item;
> +
> +		sum += sizeof(struct erofs_xattr_entry);
> +		sum += item->keylen + item->vallen;
> +		sum = EROFS_XATTR_ALIGN(sum);
> +	}
> +
> +	sum += sizeof(struct erofs_xattr_ibody_header);
> +
> +	return EROFS_XATTR_ALIGN(sum);
> +}
> +
> +char *xattr_data(struct list_head *hlist, int xattr_size)
> +{
> +	struct xattr_list *lst;
> +	char *buf, *pbuf;
> +	unsigned int size = 0;
> +	struct erofs_xattr_ibody_header header = {
> +		.h_checksum = 0,
> +		.h_shared_count = 0,
> +	};
> +
> +	erofs_info("xattr_size=%d", xattr_size);
> +	buf = malloc(xattr_size);
> +	if (!buf)
> +		return ERR_PTR(-ENOMEM);
> +
> +	memset(buf, 0, xattr_size);
> +	pbuf = buf + sizeof(struct erofs_xattr_ibody_header);
> +
> +	list_for_each_entry(lst, hlist, list) {
> +		struct erofs_xattr_entry entry;
> +		struct xattr_item *item = lst->item;
> +
> +		entry.e_name_index = item->index;
> +		entry.e_name_len = item->keylen;
> +		entry.e_value_size = cpu_to_le16(item->vallen);
> +
> +		BUG_ON(size > xattr_size);
> +		memcpy(pbuf + size, &entry, sizeof(entry));
> +
> +		size += sizeof(struct erofs_xattr_entry);
> +		memcpy(pbuf + size, item->buf, item->keylen + item->vallen);
> +		size += item->keylen + item->vallen;
> +		size = EROFS_XATTR_ALIGN(size);
> +	}
> +
> +	memcpy(buf, &header, sizeof(header));
> +
> +	return buf;
> +}
> +
> 

^ permalink raw reply	[flat|nested] 12+ messages in thread

* [PATCH] erofs-utils: introduce xattr support
  2019-08-05 14:54 ` Li Guifu
@ 2019-08-05 17:30   ` Gao Xiang
  2019-08-11 17:10     ` [PATCH v2] erofs-utils: introduce preliminary " Gao Xiang
  0 siblings, 1 reply; 12+ messages in thread
From: Gao Xiang @ 2019-08-05 17:30 UTC (permalink / raw)


Hi Guifu,

On Mon, Aug 05, 2019@10:54:52PM +0800, Li Guifu wrote:
> Hi
> It seems look good, a great step.
> A new develop branch is a googd choice to make this more stable
> and the hash function would better use a light weight one like
> full_name_hash in the kernel

Yes, you are right. it seems a great baby step of xattr.
I will make a new branch to test this of course.

BTW, Guifu, please help review this patch as well, and
don't forget to tag "Reviewed-by:" if it looks good to you.

Thanks,
Gao Xiang

> 
> ?? 2019/7/26 19:50, htyuxe+dhbrei4sq0df8 at grr.la ????:
> > load xattrs from source files and pack them into target image.
> > ---
> >   include/erofs/hashtable.h | 502 ++++++++++++++++++++++++++++++++++++++
> >   include/erofs/internal.h  |   2 +-
> >   include/erofs/xattr.h     |  22 ++
> >   lib/Makefile.am           |   3 +-
> >   lib/inode.c               |  23 ++
> >   lib/xattr.c               | 319 ++++++++++++++++++++++++
> >   6 files changed, 869 insertions(+), 2 deletions(-)
> >   create mode 100644 include/erofs/hashtable.h
> >   create mode 100644 include/erofs/xattr.h
> >   create mode 100644 lib/xattr.c
> > 
> > diff --git a/include/erofs/hashtable.h b/include/erofs/hashtable.h
> > new file mode 100644
> > index 0000000..349a655
> > --- /dev/null
> > +++ b/include/erofs/hashtable.h
> > @@ -0,0 +1,502 @@
> > +/* SPDX-License-Identifier: GPL-2.0+ */
> > +/*
> > + * include/erofs/hashtable.h
> > + *
> > + */
> > +
> > +#ifndef __EROFS_HASHTABLE_H
> > +#define __EROFS_HASHTABLE_H
> > +
> > +#define BITS_PER_LONG 32
> > +#ifndef __always_inline
> > +#define __always_inline inline
> > +#endif
> > +
> > +/* 2^31 + 2^29 - 2^25 + 2^22 - 2^19 - 2^16 + 1 */
> > +#define GOLDEN_RATIO_PRIME_32 0x9e370001UL
> > +/*  2^63 + 2^61 - 2^57 + 2^54 - 2^51 - 2^18 + 1 */
> > +#define GOLDEN_RATIO_PRIME_64 0x9e37fffffffc0001UL
> > +
> > +#if BITS_PER_LONG == 32
> > +#define GOLDEN_RATIO_PRIME GOLDEN_RATIO_PRIME_32
> > +#define hash_long(val, bits) hash_32(val, bits)
> > +#elif BITS_PER_LONG == 64
> > +#define hash_long(val, bits) hash_64(val, bits)
> > +#define GOLDEN_RATIO_PRIME GOLDEN_RATIO_PRIME_64
> > +#else
> > +#error Wordsize not 32 or 64
> > +#endif
> > +
> > +struct hlist_head {
> > +	struct hlist_node *first;
> > +};
> > +
> > +struct hlist_node {
> > +	struct hlist_node *next, **pprev;
> > +};
> > +
> > +/*
> > + * Architectures might want to move the poison pointer offset
> > + * into some well-recognized area such as 0xdead000000000000,
> > + * that is also not mappable by user-space exploits:
> > + */
> > +#ifdef CONFIG_ILLEGAL_POINTER_VALUE
> > +# define POISON_POINTER_DELTA _AC(CONFIG_ILLEGAL_POINTER_VALUE, UL)
> > +#else
> > +# define POISON_POINTER_DELTA 0
> > +#endif
> > +
> > +/*
> > + * These are non-NULL pointers that will result in page faults
> > + * under normal circumstances, used to verify that nobody uses
> > + * non-initialized list entries.
> > + */
> > +#define LIST_POISON1 ((void *) 0x00100100 + POISON_POINTER_DELTA)
> > +#define LIST_POISON2 ((void *) 0x00200200 + POISON_POINTER_DELTA)
> > +
> > +#undef offsetof
> > +#ifdef __compiler_offsetof
> > +#define offsetof(TYPE, MEMBER) __compiler_offsetof(TYPE, MEMBER)
> > +#else
> > +#define offsetof(TYPE, MEMBER) ((size_t) &((TYPE *)0)->MEMBER)
> > +#endif
> > +
> > +/*
> > + * Double linked lists with a single pointer list head.
> > + * Mostly useful for hash tables where the two pointer list head is
> > + * too wasteful.
> > + * You lose the ability to access the tail in O(1).
> > + */
> > +
> > +#define HLIST_HEAD_INIT { .first = NULL }
> > +#define HLIST_HEAD(name) struct hlist_head name = { .first = NULL }
> > +#define INIT_HLIST_HEAD(ptr) ((ptr)->first = NULL)
> > +static inline void INIT_HLIST_NODE(struct hlist_node *h)
> > +{
> > +	h->next = NULL;
> > +	h->pprev = NULL;
> > +}
> > +
> > +static inline int hlist_unhashed(const struct hlist_node *h)
> > +{
> > +	return !h->pprev;
> > +}
> > +
> > +static inline int hlist_empty(const struct hlist_head *h)
> > +{
> > +	return !h->first;
> > +}
> > +
> > +static inline void __hlist_del(struct hlist_node *n)
> > +{
> > +	struct hlist_node *next = n->next;
> > +	struct hlist_node **pprev = n->pprev;
> > +
> > +	*pprev = next;
> > +	if (next)
> > +		next->pprev = pprev;
> > +}
> > +
> > +static inline void hlist_del(struct hlist_node *n)
> > +{
> > +	__hlist_del(n);
> > +	n->next = LIST_POISON1;
> > +	n->pprev = LIST_POISON2;
> > +}
> > +
> > +static inline void hlist_del_init(struct hlist_node *n)
> > +{
> > +	if (!hlist_unhashed(n)) {
> > +		__hlist_del(n);
> > +		INIT_HLIST_NODE(n);
> > +	}
> > +}
> > +
> > +static inline void hlist_add_head(struct hlist_node *n, struct hlist_head *h)
> > +{
> > +	struct hlist_node *first = h->first;
> > +
> > +	n->next = first;
> > +	if (first)
> > +		first->pprev = &n->next;
> > +	h->first = n;
> > +	n->pprev = &h->first;
> > +}
> > +
> > +/* next must be != NULL */
> > +static inline void hlist_add_before(struct hlist_node *n,
> > +					struct hlist_node *next)
> > +{
> > +	n->pprev = next->pprev;
> > +	n->next = next;
> > +	next->pprev = &n->next;
> > +	*(n->pprev) = n;
> > +}
> > +
> > +static inline void hlist_add_behind(struct hlist_node *n,
> > +				    struct hlist_node *prev)
> > +{
> > +	n->next = prev->next;
> > +	prev->next = n;
> > +	n->pprev = &prev->next;
> > +
> > +	if (n->next)
> > +		n->next->pprev  = &n->next;
> > +}
> > +
> > +/* after that we'll appear to be on some hlist and hlist_del will work */
> > +static inline void hlist_add_fake(struct hlist_node *n)
> > +{
> > +	n->pprev = &n->next;
> > +}
> > +
> > +/*
> > + * Move a list from one list head to another. Fixup the pprev
> > + * reference of the first entry if it exists.
> > + */
> > +static inline void hlist_move_list(struct hlist_head *old,
> > +				   struct hlist_head *new)
> > +{
> > +	new->first = old->first;
> > +	if (new->first)
> > +		new->first->pprev = &new->first;
> > +	old->first = NULL;
> > +}
> > +
> > +#define hlist_entry(ptr, type, member) container_of(ptr, type, member)
> > +
> > +#define hlist_for_each(pos, head) \
> > +	for (pos = (head)->first; pos; pos = pos->next)
> > +
> > +#define hlist_for_each_safe(pos, n, head) \
> > +	for (pos = (head)->first; pos && ({ n = pos->next; 1; }); \
> > +	     pos = n)
> > +
> > +#define hlist_entry_safe(ptr, type, member) \
> > +	({ typeof(ptr) ____ptr = (ptr); \
> > +	   ____ptr ? hlist_entry(____ptr, type, member) : NULL; \
> > +	})
> > +
> > +/**
> > + * hlist_for_each_entry	- iterate over list of given type
> > + * @pos:the type * to use as a loop cursor.
> > + * @head:the head for your list.
> > + * @member:the name of the hlist_node within the struct.
> > + */
> > +#define hlist_for_each_entry(pos, head, member)				\
> > +	for (pos = hlist_entry_safe((head)->first, typeof(*(pos)), member);\
> > +	     pos;							\
> > +	     pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member))
> > +
> > +/**
> > + * hlist_for_each_entry_continue
> > + * iterate over a hlist continuing after current point
> > + * @pos:the type * to use as a loop cursor.
> > + * @member:the name of the hlist_node within the struct.
> > + */
> > +#define hlist_for_each_entry_continue(pos, member)			\
> > +	for (pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member);\
> > +	     pos;							\
> > +	     pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member))
> > +
> > +/**
> > + * hlist_for_each_entry_from
> > + * iterate over a hlist continuing from current point
> > + * @pos:	the type * to use as a loop cursor.
> > + * @member:	the name of the hlist_node within the struct.
> > + */
> > +#define hlist_for_each_entry_from(pos, member)				\
> > +	for (; pos;							\
> > +	     pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member))
> > +
> > +/**
> > + * hlist_for_each_entry_safe
> > + * iterate over list of given type safe against removal of list entry
> > + * @pos:the type * to use as a loop cursor.
> > + * @n:another &struct hlist_node to use as temporary storage
> > + * @head:the head for your list.
> > + * @member:the name of the hlist_node within the struct.
> > + */
> > +#define hlist_for_each_entry_safe(pos, n, head, member) 		\
> > +	for (pos = hlist_entry_safe((head)->first, typeof(*pos), member);\
> > +		pos && ({ n = pos->member.next; 1; });			\
> > +		pos = hlist_entry_safe(n, typeof(*pos), member))
> > +
> > +static __always_inline u64 hash_64(u64 val, unsigned int bits)
> > +{
> > +	u64 hash = val;
> > +
> > +#if defined(CONFIG_ARCH_HAS_FAST_MULTIPLIER) && BITS_PER_LONG == 64
> > +	hash = hash * GOLDEN_RATIO_PRIME_64;
> > +#else
> > +	/*  Sigh, gcc can't optimise this alone like it does for 32 bits. */
> > +	u64 n = hash;
> > +
> > +	n <<= 18;
> > +	hash -= n;
> > +	n <<= 33;
> > +	hash -= n;
> > +	n <<= 3;
> > +	hash += n;
> > +	n <<= 3;
> > +	hash -= n;
> > +	n <<= 4;
> > +	hash += n;
> > +	n <<= 2;
> > +	hash += n;
> > +#endif
> > +
> > +	/* High bits are more random, so use them. */
> > +	return hash >> (64 - bits);
> > +}
> > +
> > +static inline u32 hash_32(u32 val, unsigned int bits)
> > +{
> > +	/* On some cpus multiply is faster, on others gcc will do shifts */
> > +	u32 hash = val * GOLDEN_RATIO_PRIME_32;
> > +
> > +	/* High bits are more random, so use them. */
> > +	return hash >> (32 - bits);
> > +}
> > +
> > +/**
> > + * ilog2 - log of base 2 of 32-bit or a 64-bit unsigned value
> > + * @n - parameter
> > + *
> > + * constant-capable log of base 2 calculation
> > + * - this can be used to initialise global variables from constant data, hence
> > + *   the massive ternary operator construction
> > + *
> > + * selects the appropriately-sized optimised version depending on sizeof(n)
> > + */
> > +#define ilog2(n)				\
> > +(								\
> > +	(n) & (1ULL << 63) ? 63 :	\
> > +	(n) & (1ULL << 62) ? 62 :	\
> > +	(n) & (1ULL << 61) ? 61 :	\
> > +	(n) & (1ULL << 60) ? 60 :	\
> > +	(n) & (1ULL << 59) ? 59 :	\
> > +	(n) & (1ULL << 58) ? 58 :	\
> > +	(n) & (1ULL << 57) ? 57 :	\
> > +	(n) & (1ULL << 56) ? 56 :	\
> > +	(n) & (1ULL << 55) ? 55 :	\
> > +	(n) & (1ULL << 54) ? 54 :	\
> > +	(n) & (1ULL << 53) ? 53 :	\
> > +	(n) & (1ULL << 52) ? 52 :	\
> > +	(n) & (1ULL << 51) ? 51 :	\
> > +	(n) & (1ULL << 50) ? 50 :	\
> > +	(n) & (1ULL << 49) ? 49 :	\
> > +	(n) & (1ULL << 48) ? 48 :	\
> > +	(n) & (1ULL << 47) ? 47 :	\
> > +	(n) & (1ULL << 46) ? 46 :	\
> > +	(n) & (1ULL << 45) ? 45 :	\
> > +	(n) & (1ULL << 44) ? 44 :	\
> > +	(n) & (1ULL << 43) ? 43 :	\
> > +	(n) & (1ULL << 42) ? 42 :	\
> > +	(n) & (1ULL << 41) ? 41 :	\
> > +	(n) & (1ULL << 40) ? 40 :	\
> > +	(n) & (1ULL << 39) ? 39 :	\
> > +	(n) & (1ULL << 38) ? 38 :	\
> > +	(n) & (1ULL << 37) ? 37 :	\
> > +	(n) & (1ULL << 36) ? 36 :	\
> > +	(n) & (1ULL << 35) ? 35 :	\
> > +	(n) & (1ULL << 34) ? 34 :	\
> > +	(n) & (1ULL << 33) ? 33 :	\
> > +	(n) & (1ULL << 32) ? 32 :	\
> > +	(n) & (1ULL << 31) ? 31 :	\
> > +	(n) & (1ULL << 30) ? 30 :	\
> > +	(n) & (1ULL << 29) ? 29 :	\
> > +	(n) & (1ULL << 28) ? 28 :	\
> > +	(n) & (1ULL << 27) ? 27 :	\
> > +	(n) & (1ULL << 26) ? 26 :	\
> > +	(n) & (1ULL << 25) ? 25 :	\
> > +	(n) & (1ULL << 24) ? 24 :	\
> > +	(n) & (1ULL << 23) ? 23 :	\
> > +	(n) & (1ULL << 22) ? 22 :	\
> > +	(n) & (1ULL << 21) ? 21 :	\
> > +	(n) & (1ULL << 20) ? 20 :	\
> > +	(n) & (1ULL << 19) ? 19 :	\
> > +	(n) & (1ULL << 18) ? 18 :	\
> > +	(n) & (1ULL << 17) ? 17 :	\
> > +	(n) & (1ULL << 16) ? 16 :	\
> > +	(n) & (1ULL << 15) ? 15 :	\
> > +	(n) & (1ULL << 14) ? 14 :	\
> > +	(n) & (1ULL << 13) ? 13 :	\
> > +	(n) & (1ULL << 12) ? 12 :	\
> > +	(n) & (1ULL << 11) ? 11 :	\
> > +	(n) & (1ULL << 10) ? 10 :	\
> > +	(n) & (1ULL <<  9) ?  9 :	\
> > +	(n) & (1ULL <<  8) ?  8 :	\
> > +	(n) & (1ULL <<  7) ?  7 :	\
> > +	(n) & (1ULL <<  6) ?  6 :	\
> > +	(n) & (1ULL <<  5) ?  5 :	\
> > +	(n) & (1ULL <<  4) ?  4 :	\
> > +	(n) & (1ULL <<  3) ?  3 :	\
> > +	(n) & (1ULL <<  2) ?  2 :	\
> > +	(n) & (1ULL <<  1) ?  1 : 0	\
> > +)
> > +
> > +static const uint16_t crc16tab[256] = {
> > +	0x0000, 0x1021, 0x2042, 0x3063, 0x4084, 0x50a5, 0x60c6, 0x70e7,
> > +	0x8108, 0x9129, 0xa14a, 0xb16b, 0xc18c, 0xd1ad, 0xe1ce, 0xf1ef,
> > +	0x1231, 0x0210, 0x3273, 0x2252, 0x52b5, 0x4294, 0x72f7, 0x62d6,
> > +	0x9339, 0x8318, 0xb37b, 0xa35a, 0xd3bd, 0xc39c, 0xf3ff, 0xe3de,
> > +	0x2462, 0x3443, 0x0420, 0x1401, 0x64e6, 0x74c7, 0x44a4, 0x5485,
> > +	0xa56a, 0xb54b, 0x8528, 0x9509, 0xe5ee, 0xf5cf, 0xc5ac, 0xd58d,
> > +	0x3653, 0x2672, 0x1611, 0x0630, 0x76d7, 0x66f6, 0x5695, 0x46b4,
> > +	0xb75b, 0xa77a, 0x9719, 0x8738, 0xf7df, 0xe7fe, 0xd79d, 0xc7bc,
> > +	0x48c4, 0x58e5, 0x6886, 0x78a7, 0x0840, 0x1861, 0x2802, 0x3823,
> > +	0xc9cc, 0xd9ed, 0xe98e, 0xf9af, 0x8948, 0x9969, 0xa90a, 0xb92b,
> > +	0x5af5, 0x4ad4, 0x7ab7, 0x6a96, 0x1a71, 0x0a50, 0x3a33, 0x2a12,
> > +	0xdbfd, 0xcbdc, 0xfbbf, 0xeb9e, 0x9b79, 0x8b58, 0xbb3b, 0xab1a,
> > +	0x6ca6, 0x7c87, 0x4ce4, 0x5cc5, 0x2c22, 0x3c03, 0x0c60, 0x1c41,
> > +	0xedae, 0xfd8f, 0xcdec, 0xddcd, 0xad2a, 0xbd0b, 0x8d68, 0x9d49,
> > +	0x7e97, 0x6eb6, 0x5ed5, 0x4ef4, 0x3e13, 0x2e32, 0x1e51, 0x0e70,
> > +	0xff9f, 0xefbe, 0xdfdd, 0xcffc, 0xbf1b, 0xaf3a, 0x9f59, 0x8f78,
> > +	0x9188, 0x81a9, 0xb1ca, 0xa1eb, 0xd10c, 0xc12d, 0xf14e, 0xe16f,
> > +	0x1080, 0x00a1, 0x30c2, 0x20e3, 0x5004, 0x4025, 0x7046, 0x6067,
> > +	0x83b9, 0x9398, 0xa3fb, 0xb3da, 0xc33d, 0xd31c, 0xe37f, 0xf35e,
> > +	0x02b1, 0x1290, 0x22f3, 0x32d2, 0x4235, 0x5214, 0x6277, 0x7256,
> > +	0xb5ea, 0xa5cb, 0x95a8, 0x8589, 0xf56e, 0xe54f, 0xd52c, 0xc50d,
> > +	0x34e2, 0x24c3, 0x14a0, 0x0481, 0x7466, 0x6447, 0x5424, 0x4405,
> > +	0xa7db, 0xb7fa, 0x8799, 0x97b8, 0xe75f, 0xf77e, 0xc71d, 0xd73c,
> > +	0x26d3, 0x36f2, 0x0691, 0x16b0, 0x6657, 0x7676, 0x4615, 0x5634,
> > +	0xd94c, 0xc96d, 0xf90e, 0xe92f, 0x99c8, 0x89e9, 0xb98a, 0xa9ab,
> > +	0x5844, 0x4865, 0x7806, 0x6827, 0x18c0, 0x08e1, 0x3882, 0x28a3,
> > +	0xcb7d, 0xdb5c, 0xeb3f, 0xfb1e, 0x8bf9, 0x9bd8, 0xabbb, 0xbb9a,
> > +	0x4a75, 0x5a54, 0x6a37, 0x7a16, 0x0af1, 0x1ad0, 0x2ab3, 0x3a92,
> > +	0xfd2e, 0xed0f, 0xdd6c, 0xcd4d, 0xbdaa, 0xad8b, 0x9de8, 0x8dc9,
> > +	0x7c26, 0x6c07, 0x5c64, 0x4c45, 0x3ca2, 0x2c83, 0x1ce0, 0x0cc1,
> > +	0xef1f, 0xff3e, 0xcf5d, 0xdf7c, 0xaf9b, 0xbfba, 0x8fd9, 0x9ff8,
> > +	0x6e17, 0x7e36, 0x4e55, 0x5e74, 0x2e93, 0x3eb2, 0x0ed1, 0x1ef0
> > +};
> > +
> > +uint16_t crc16(const char *buf, int len)
> > +{
> > +	int counter;
> > +	uint16_t crc = 0;
> > +
> > +	for (counter = 0; counter < len; counter++)
> > +		crc = (crc<<8) ^ crc16tab[((crc>>8) ^ *buf++)&0x00FF];
> > +	return crc;
> > +}
> > +
> > +#define DEFINE_HASHTABLE(name, bits)					\
> > +	struct hlist_head name[1 << (bits)] =				\
> > +			{ [0 ... ((1 << (bits)) - 1)] = HLIST_HEAD_INIT }
> > +
> > +#define DECLARE_HASHTABLE(name, bits)					\
> > +	struct hlist_head name[1 << (bits)]
> > +
> > +#define HASH_SIZE(name) (ARRAY_SIZE(name))
> > +#define HASH_BITS(name) ilog2(HASH_SIZE(name))
> > +
> > +/* Use hash_32 when possible to allow for fast 32bit hashing in 64bit kernels*/
> > +#define hash_min(val, bits)						\
> > +	(sizeof(val) <= 4 ? hash_32(val, bits) : hash_long(val, bits))
> > +
> > +static inline void __hash_init(struct hlist_head *ht, unsigned int sz)
> > +{
> > +	unsigned int i;
> > +
> > +	for (i = 0; i < sz; i++)
> > +		INIT_HLIST_HEAD(&ht[i]);
> > +}
> > +
> > +/**
> > + * hash_init - initialize a hash table
> > + * @hashtable: hashtable to be initialized
> > + *
> > + * Calculates the size of the hashtable from the given parameter, otherwise
> > + * same as hash_init_size.
> > + *
> > + * This has to be a macro since HASH_BITS() will not work on pointers since
> > + * it calculates the size during preprocessing.
> > + */
> > +#define hash_init(hashtable) __hash_init(hashtable, HASH_SIZE(hashtable))
> > +
> > +/**
> > + * hash_add - add an object to a hashtable
> > + * @hashtable: hashtable to add to
> > + * @node: the &struct hlist_node of the object to be added
> > + * @key: the key of the object to be added
> > + */
> > +#define hash_add(hashtable, node, key)					\
> > +	hlist_add_head(node, &hashtable[hash_min(key, HASH_BITS(hashtable))])
> > +
> > +/**
> > + * hash_hashed - check whether an object is in any hashtable
> > + * @node: the &struct hlist_node of the object to be checked
> > + */
> > +static inline bool hash_hashed(struct hlist_node *node)
> > +{
> > +	return !hlist_unhashed(node);
> > +}
> > +
> > +static inline bool __hash_empty(struct hlist_head *ht, unsigned int sz)
> > +{
> > +	unsigned int i;
> > +
> > +	for (i = 0; i < sz; i++)
> > +		if (!hlist_empty(&ht[i]))
> > +			return false;
> > +
> > +	return true;
> > +}
> > +
> > +/**
> > + * hash_empty - check whether a hashtable is empty
> > + * @hashtable: hashtable to check
> > + *
> > + * This has to be a macro since HASH_BITS() will not work on pointers since
> > + * it calculates the size during preprocessing.
> > + */
> > +#define hash_empty(hashtable) __hash_empty(hashtable, HASH_SIZE(hashtable))
> > +
> > +/**
> > + * hash_del - remove an object from a hashtable
> > + * @node: &struct hlist_node of the object to remove
> > + */
> > +static inline void hash_del(struct hlist_node *node)
> > +{
> > +	hlist_del_init(node);
> > +}
> > +
> > +/**
> > + * hash_for_each - iterate over a hashtable
> > + * @name: hashtable to iterate
> > + * @bkt: integer to use as bucket loop cursor
> > + * @obj: the type * to use as a loop cursor for each entry
> > + * @member: the name of the hlist_node within the struct
> > + */
> > +#define hash_for_each(name, bkt, obj, member)				\
> > +	for ((bkt) = 0, obj = NULL; obj == NULL && (bkt) < HASH_SIZE(name);\
> > +			(bkt)++)\
> > +		hlist_for_each_entry(obj, &name[bkt], member)
> > +
> > +/**
> > + * hash_for_each_safe - iterate over a hashtable safe against removal of
> > + * hash entry
> > + * @name: hashtable to iterate
> > + * @bkt: integer to use as bucket loop cursor
> > + * @tmp: a &struct used for temporary storage
> > + * @obj: the type * to use as a loop cursor for each entry
> > + * @member: the name of the hlist_node within the struct
> > + */
> > +#define hash_for_each_safe(name, bkt, tmp, obj, member)			\
> > +	for ((bkt) = 0, obj = NULL; obj == NULL && (bkt) < HASH_SIZE(name);\
> > +			(bkt)++)\
> > +		hlist_for_each_entry_safe(obj, tmp, &name[bkt], member)
> > +
> > +/**
> > + * hash_for_each_possible - iterate over all possible objects hashing to the
> > + * same bucket
> > + * @name: hashtable to iterate
> > + * @obj: the type * to use as a loop cursor for each entry
> > + * @member: the name of the hlist_node within the struct
> > + * @key: the key of the objects to iterate over
> > + */
> > +#define hash_for_each_possible(name, obj, member, key)			\
> > +	hlist_for_each_entry(obj, &name[hash_min(key, HASH_BITS(name))], member)
> > +
> > +#endif
> > diff --git a/include/erofs/internal.h b/include/erofs/internal.h
> > index b7ce6f8..33a72b5 100644
> > --- a/include/erofs/internal.h
> > +++ b/include/erofs/internal.h
> > @@ -59,7 +59,7 @@ struct erofs_sb_info {
> >   extern struct erofs_sb_info sbi;
> >   struct erofs_inode {
> > -	struct list_head i_hash, i_subdirs;
> > +	struct list_head i_hash, i_subdirs, i_xattrs;
> >   	unsigned int i_count;
> >   	struct erofs_inode *i_parent;
> > diff --git a/include/erofs/xattr.h b/include/erofs/xattr.h
> > new file mode 100644
> > index 0000000..dff9fd6
> > --- /dev/null
> > +++ b/include/erofs/xattr.h
> > @@ -0,0 +1,22 @@
> > +// SPDX-License-Identifier: GPL-2.0+
> > +/*
> > + * include/erofs/xattr.h
> > + *
> > + */
> > +
> > +#ifndef __EROFS_XATTR_H
> > +#define __EROFS_XATTR_H
> > +
> > +#define XATTR_COUNT(_size)	({\
> > +	u32 __size = le16_to_cpu(_size); \
> > +	((__size) == 0) ? 0 : \
> > +	(_size - sizeof(struct erofs_xattr_ibody_header)) / \
> > +	sizeof(struct erofs_xattr_entry) + 1; })
> > +
> > +
> > +int cust_xattr(struct list_head *hlist);
> > +int read_xattr_from_src(const char *path, struct list_head *hlist);
> > +int xattr_entry_size(struct list_head *hlist);
> > +char *xattr_data(struct list_head *hlist, int size);
> > +
> > +#endif
> > diff --git a/lib/Makefile.am b/lib/Makefile.am
> > index dea82f7..cbe3243 100644
> > --- a/lib/Makefile.am
> > +++ b/lib/Makefile.am
> > @@ -2,7 +2,8 @@
> >   # Makefile.am
> >   noinst_LTLIBRARIES = liberofs.la
> > -liberofs_la_SOURCES = config.c io.c cache.c inode.c compress.c compressor.c
> > +liberofs_la_SOURCES = config.c io.c cache.c inode.c compress.c compressor.c \
> > +		      xattr.c
> >   liberofs_la_CFLAGS = -Wall -Werror -I$(top_srcdir)/include
> >   if ENABLE_LZ4
> >   liberofs_la_CFLAGS += ${LZ4_CFLAGS}
> > diff --git a/lib/inode.c b/lib/inode.c
> > index 8b38270..615f117 100644
> > --- a/lib/inode.c
> > +++ b/lib/inode.c
> > @@ -18,6 +18,7 @@
> >   #include "erofs/cache.h"
> >   #include "erofs/io.h"
> >   #include "erofs/compress.h"
> > +#include "erofs/xattr.h"
> >   struct erofs_sb_info sbi;
> > @@ -364,8 +365,10 @@ static bool erofs_bh_flush_write_inode(struct erofs_buffer_head *bh)
> >   	/* let's support v1 currently */
> >   	struct erofs_inode_v1 v1 = {0};
> >   	int ret;
> > +	uint16_t count = XATTR_COUNT(inode->xattr_isize);
> >   	v1.i_advise = cpu_to_le16(0 | (inode->data_mapping_mode << 1));
> > +	v1.i_xattr_icount = cpu_to_le16(count);
> >   	v1.i_mode = cpu_to_le16(inode->i_mode);
> >   	v1.i_nlink = cpu_to_le16(inode->i_nlink);
> >   	v1.i_size = cpu_to_le32((u32)inode->i_size);
> > @@ -399,6 +402,20 @@ static bool erofs_bh_flush_write_inode(struct erofs_buffer_head *bh)
> >   		return false;
> >   	off += inode->inode_isize;
> > +	if (inode->xattr_isize) {
> > +		char *pbuf = xattr_data(&inode->i_xattrs, inode->xattr_isize);
> > +
> > +		if (IS_ERR(pbuf))
> > +			return false;
> > +
> > +		ret = dev_write(pbuf, off, inode->xattr_isize);
> > +		free(pbuf);
> > +		if (ret)
> > +			return false;
> > +
> > +		off += inode->xattr_isize;
> > +	}
> > +
> >   	if (inode->extent_isize) {
> >   		/* write compression metadata */
> >   		off = Z_EROFS_VLE_EXTENT_ALIGN(off);
> > @@ -452,6 +469,7 @@ int erofs_prepare_inode_buffer(struct erofs_inode *inode)
> >   	DBG_BUGON(inode->bh || inode->bh_inline);
> > +	inode->xattr_isize = xattr_entry_size(&inode->i_xattrs);
> >   	inodesize = inode->inode_isize + inode->xattr_isize +
> >   		    inode->extent_isize;
> > @@ -612,6 +630,7 @@ struct erofs_inode *erofs_new_inode(void)
> >   	inode->i_count = 1;
> >   	init_list_head(&inode->i_subdirs);
> > +	init_list_head(&inode->i_xattrs);
> >   	inode->xattr_isize = 0;
> >   	inode->extent_isize = 0;
> > @@ -699,6 +718,10 @@ struct erofs_inode *erofs_mkfs_build_tree(struct erofs_inode *dir)
> >   	struct dirent *dp;
> >   	struct erofs_dentry *d;
> > +	ret = read_xattr_from_src(dir->i_srcpath, &dir->i_xattrs);
> > +	if (ret)
> > +		return ERR_PTR(ret);
> > +
> >   	if (!S_ISDIR(dir->i_mode)) {
> >   		if (S_ISLNK(dir->i_mode)) {
> >   			char *const symlink = malloc(dir->i_size);
> > diff --git a/lib/xattr.c b/lib/xattr.c
> > new file mode 100644
> > index 0000000..6278abc
> > --- /dev/null
> > +++ b/lib/xattr.c
> > @@ -0,0 +1,319 @@
> > +// SPDX-License-Identifier: GPL-2.0+
> > +/*
> > + * lib/xattr.c
> > + *
> > + */
> > +
> > +#include <stdlib.h>
> > +#include <sys/types.h>
> > +#include <sys/xattr.h>
> > +#include <linux/xattr.h>
> > +#include <errno.h>
> > +#include <string.h>
> > +
> > +#include "erofs/defs.h"
> > +#include "erofs/print.h"
> > +#include "erofs/list.h"
> > +#include "erofs/internal.h"
> > +#include "erofs/hashtable.h"
> > +#include "err.h"
> > +#define EROFS_XATTR_HASH_TABLE_BITS 16
> > +
> > +struct xattr_item {
> > +	const char *buf;
> > +	unsigned int keylen;
> > +	unsigned int vallen;
> > +	unsigned int count;
> > +	u8 index;
> > +	struct hlist_node node;
> > +};
> > +
> > +struct xattr_list {
> > +	struct xattr_item *item;
> > +	struct list_head list;
> > +};
> > +
> > +DECLARE_HASHTABLE(my_hash_table, EROFS_XATTR_HASH_TABLE_BITS);
> > +
> > +struct xattr_prefix {
> > +	const char *prefix;
> > +	uint16_t prefix_len;
> > +	u8 index;
> > +} prefix[] = {
> > +	{
> > +		XATTR_USER_PREFIX,
> > +		XATTR_USER_PREFIX_LEN,
> > +		EROFS_XATTR_INDEX_USER
> > +	},
> > +	{
> > +		XATTR_SECURITY_PREFIX,
> > +		XATTR_SYSTEM_PREFIX_LEN,
> > +		EROFS_XATTR_INDEX_SECURITY
> > +	},
> > +	{
> > +		XATTR_TRUSTED_PREFIX,
> > +		XATTR_TRUSTED_PREFIX_LEN,
> > +		EROFS_XATTR_INDEX_TRUSTED
> > +	},
> > +	{NULL, 0, 0},
> > +};
> > +
> > +static inline void hxattr_add(struct hlist_node *node, uint16_t key)
> > +{
> > +	hash_add(my_hash_table, node, key);
> > +}
> > +
> > +static inline void hxattr_del(struct hlist_node *node)
> > +{
> > +	hash_del(node);
> > +}
> > +
> > +static struct xattr_item *hxattr_match(const char *buf, int len, u8 index)
> > +{
> > +	struct xattr_item *item;
> > +	uint16_t mkey = crc16(buf, len);
> > +
> > +	hash_for_each_possible(my_hash_table, item, node, mkey) {
> > +		if (index == item->index &&
> > +		    len == (item->keylen + item->vallen) &&
> > +		    !memcmp(buf, item->buf, len)) {
> > +			return item;
> > +		}
> > +	}
> > +
> > +	return ERR_PTR(-ENOENT);
> > +}
> > +
> > +static bool match_index(const char *key, u8 *index, uint16_t *len)
> > +{
> > +	struct xattr_prefix *p = prefix;
> > +
> > +	while (p->prefix) {
> > +		if (strncmp(p->prefix, key, p->prefix_len)) {
> > +			*len = p->prefix_len;
> > +			*index = p->index;
> > +			return true;
> > +		}
> > +		p++;
> > +	}
> > +
> > +	return false;
> > +}
> > +
> > +static struct xattr_item *new_xattr(const char *buf, u8 index,
> > +				    int keylen, int vallen)
> > +{
> > +	struct xattr_item *item = malloc(sizeof(*item));
> > +
> > +	if (!item)
> > +		return ERR_PTR(-ENOMEM);
> > +
> > +	memset(item, 0, sizeof(*item));
> > +	INIT_HLIST_NODE(&item->node);
> > +	item->buf = buf;
> > +	item->keylen = keylen;
> > +	item->vallen = vallen;
> > +	item->count = 1;
> > +	item->index = index;
> > +	if (!item->index)
> > +		return ERR_PTR(-EPERM);
> > +
> > +	return item;
> > +}
> > +
> > +static int xattr_add(struct list_head *hlist, struct xattr_item *item)
> > +{
> > +	struct xattr_list *mlist = malloc(sizeof(*mlist));
> > +
> > +	if (!mlist)
> > +		return -ENOMEM;
> > +
> > +	init_list_head(&mlist->list);
> > +	mlist->item = item;
> > +	list_add(&mlist->list, hlist);
> > +	return 0;
> > +}
> > +
> > +static struct xattr_item *list_xattr_value(const char *path, const char *key)
> > +{
> > +	ssize_t keylen, vallen;
> > +	char *kxattr;
> > +	struct xattr_item *item;
> > +	u8 index;
> > +	uint16_t prelen, suflen;
> > +
> > +	/* Output attribute key.*/
> > +	erofs_info("path:%s key: [%s] ", path, key);
> > +
> > +	keylen = strlen(key);
> > +	if (!match_index(key, &index, &prelen))
> > +		return ERR_PTR(-ENODATA);
> > +
> > +	BUG_ON(keylen < prelen);
> > +	/* Determine length of the value.*/
> > +	vallen = lgetxattr(path, key, NULL, 0);
> > +	if (vallen == -1)
> > +		return ERR_PTR(-errno);
> > +
> > +	/*
> > +	 * Allocate value buffer.
> > +	 * One extra byte is needed to append 0x00.
> > +	 */
> > +	suflen = keylen - prelen;
> > +	kxattr = malloc(suflen + vallen + 1);
> > +	if (!kxattr)
> > +		return ERR_PTR(-ENOMEM);
> > +
> > +	if (vallen == 0)
> > +		goto value_0;
> > +
> > +	/* Copy value to buffer.*/
> > +	vallen = lgetxattr(path, key, kxattr + suflen, vallen);
> > +	if (vallen == -1) {
> > +		free(kxattr);
> > +		return ERR_PTR(-errno);
> > +	}
> > +
> > +value_0:
> > +	memcpy(kxattr, key + prelen, suflen);
> > +	/* Output attribute value.*/
> > +	kxattr[suflen + vallen] = '\0';
> > +	erofs_info("value: [%s]", kxattr + suflen);
> > +
> > +	/* kxattr is used at xattr_add(), neednt free if SUCCESS */
> > +	item = hxattr_match(kxattr, suflen + vallen, index);
> > +	if (!IS_ERR(item)) {
> > +		item->count++;
> > +		free(kxattr);
> > +		return item;
> > +	}
> > +
> > +	item = new_xattr(kxattr, index, suflen, vallen);
> > +	if (IS_ERR(item))
> > +		free(kxattr);
> > +
> > +	return item;
> > +}
> > +
> > +int read_xattr_from_src(const char *path, struct list_head *hlist)
> > +{
> > +	int ret = 0;
> > +	char *kbuf, *key;
> > +	ssize_t buflen = llistxattr(path, NULL, 0);
> > +
> > +	if (buflen == -1)
> > +		return -errno;
> > +	else if (buflen == 0)
> > +		return 0;
> > +
> > +	/* Allocate the buffer.*/
> > +	kbuf = malloc(buflen);
> > +	if (!kbuf)
> > +		return -errno;
> > +
> > +	/* Copy the list of attribute keys to the buffer.*/
> > +	buflen = llistxattr(path, kbuf, buflen);
> > +	if (buflen == -1) {
> > +		ret = -errno;
> > +		goto exit_err;
> > +	}
> > +
> > +	/*
> > +	 * Loop over the list of zero terminated strings with the
> > +	 * attribute keys. Use the remaining buffer length to determine
> > +	 * the end of the list.
> > +	 */
> > +	key = kbuf;
> > +	while (buflen > 0) {
> > +		size_t keylen = strlen(key) + 1;
> > +		struct xattr_item *item = list_xattr_value(path, key);
> > +
> > +		if (!item) {
> > +			ret = -errno;
> > +			goto exit_err;
> > +		}
> > +
> > +		if (!hash_hashed(&item->node)) {
> > +			uint16_t mkey;
> > +
> > +			mkey = crc16(item->buf, item->keylen + item->vallen);
> > +			hxattr_add(&item->node, mkey);
> > +		}
> > +
> > +		if (hlist) {
> > +			ret = xattr_add(hlist, item);
> > +			if (ret < 0)
> > +				goto exit_err;
> > +		}
> > +
> > +		buflen -= keylen;
> > +		key += keylen;
> > +	}
> > +
> > +exit_err:
> > +	free(kbuf);
> > +	return ret;
> > +
> > +}
> > +
> > +int xattr_entry_size(struct list_head *hlist)
> > +{
> > +	int sum = 0;
> > +	struct xattr_list *lst;
> > +
> > +	if (list_empty(hlist))
> > +		return 0;
> > +
> > +	list_for_each_entry(lst, hlist, list) {
> > +		struct xattr_item *item = lst->item;
> > +
> > +		sum += sizeof(struct erofs_xattr_entry);
> > +		sum += item->keylen + item->vallen;
> > +		sum = EROFS_XATTR_ALIGN(sum);
> > +	}
> > +
> > +	sum += sizeof(struct erofs_xattr_ibody_header);
> > +
> > +	return EROFS_XATTR_ALIGN(sum);
> > +}
> > +
> > +char *xattr_data(struct list_head *hlist, int xattr_size)
> > +{
> > +	struct xattr_list *lst;
> > +	char *buf, *pbuf;
> > +	unsigned int size = 0;
> > +	struct erofs_xattr_ibody_header header = {
> > +		.h_checksum = 0,
> > +		.h_shared_count = 0,
> > +	};
> > +
> > +	erofs_info("xattr_size=%d", xattr_size);
> > +	buf = malloc(xattr_size);
> > +	if (!buf)
> > +		return ERR_PTR(-ENOMEM);
> > +
> > +	memset(buf, 0, xattr_size);
> > +	pbuf = buf + sizeof(struct erofs_xattr_ibody_header);
> > +
> > +	list_for_each_entry(lst, hlist, list) {
> > +		struct erofs_xattr_entry entry;
> > +		struct xattr_item *item = lst->item;
> > +
> > +		entry.e_name_index = item->index;
> > +		entry.e_name_len = item->keylen;
> > +		entry.e_value_size = cpu_to_le16(item->vallen);
> > +
> > +		BUG_ON(size > xattr_size);
> > +		memcpy(pbuf + size, &entry, sizeof(entry));
> > +
> > +		size += sizeof(struct erofs_xattr_entry);
> > +		memcpy(pbuf + size, item->buf, item->keylen + item->vallen);
> > +		size += item->keylen + item->vallen;
> > +		size = EROFS_XATTR_ALIGN(size);
> > +	}
> > +
> > +	memcpy(buf, &header, sizeof(header));
> > +
> > +	return buf;
> > +}
> > +
> > 

^ permalink raw reply	[flat|nested] 12+ messages in thread

* [PATCH v2] erofs-utils: introduce preliminary xattr support
  2019-08-05 17:30   ` Gao Xiang
@ 2019-08-11 17:10     ` Gao Xiang
  2019-10-05 14:20       ` [PATCH v3 1/2] erofs-utils: introduce inline " Gao Xiang via Linux-erofs
  0 siblings, 1 reply; 12+ messages in thread
From: Gao Xiang @ 2019-08-11 17:10 UTC (permalink / raw)


From: "htyuxe+dhbrei4sq0df8@grr.la" <htyuxe+dhbrei4sq0df8@grr.la>

Load xattrs from source files and pack them into target image.

Signed-off-by: htyuxe+dhbrei4sq0df8 at grr.la <htyuxe+dhbrei4sq0df8 at grr.la>
Signed-off-by: Li Guifu <blucerlee at gmail.com>
Signed-off-by: Gao Xiang <xiang at kernel.org>
---

This is a modified version of xattr implementation from anonymous person.

Guifu, please also take some time to test the functionality of this patch,
       therefore I can merge this to dev branch ASAP.

 configure.ac              |   1 +
 include/erofs/defs.h      |   6 +
 include/erofs/hashtable.h | 451 ++++++++++++++++++++++++++++++++++++++
 include/erofs/internal.h  |   2 +-
 include/erofs/xattr.h     |  44 ++++
 lib/Makefile.am           |   3 +-
 lib/inode.c               |  23 ++
 lib/xattr.c               | 289 ++++++++++++++++++++++++
 8 files changed, 817 insertions(+), 2 deletions(-)
 create mode 100644 include/erofs/hashtable.h
 create mode 100644 include/erofs/xattr.h
 create mode 100644 lib/xattr.c

diff --git a/configure.ac b/configure.ac
index fcdf30a..9174711 100644
--- a/configure.ac
+++ b/configure.ac
@@ -75,6 +75,7 @@ AC_CHECK_HEADERS(m4_flatten([
 	linux/falloc.h
 	linux/fs.h
 	linux/types.h
+	linux/xattr.h
 	limits.h
 	stddef.h
 	stdint.h
diff --git a/include/erofs/defs.h b/include/erofs/defs.h
index 0d9910c..b325d01 100644
--- a/include/erofs/defs.h
+++ b/include/erofs/defs.h
@@ -147,6 +147,12 @@ typedef int64_t         s64;
 #define BITS_PER_BYTE       8
 #define BITS_TO_LONGS(nr)   DIV_ROUND_UP(nr, BITS_PER_BYTE * sizeof(long))
 
+#ifdef __SIZEOF_LONG__
+#define BITS_PER_LONG (__CHAR_BIT__ * __SIZEOF_LONG__)
+#else
+#define BITS_PER_LONG __WORDSIZE
+#endif
+
 #define BUG_ON(cond)        assert(!(cond))
 
 #ifdef NDEBUG
diff --git a/include/erofs/hashtable.h b/include/erofs/hashtable.h
new file mode 100644
index 0000000..231ce99
--- /dev/null
+++ b/include/erofs/hashtable.h
@@ -0,0 +1,451 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * erofs_utils/include/erofs/hashtable.h
+ * (code originally taken from include/linux/hash{,table}.h)
+ */
+#ifndef __EROFS_HASHTABLE_H
+#define __EROFS_HASHTABLE_H
+
+#include "defs.h"
+
+/*
+ * The "GOLDEN_RATIO_PRIME" is used in ifs/btrfs/brtfs_inode.h and
+ * fs/inode.c.  It's not actually prime any more (the previous primes
+ * were actively bad for hashing), but the name remains.
+ */
+#if BITS_PER_LONG == 32
+#define GOLDEN_RATIO_PRIME GOLDEN_RATIO_32
+#define hash_long(val, bits) hash_32(val, bits)
+#elif BITS_PER_LONG == 64
+#define hash_long(val, bits) hash_64(val, bits)
+#define GOLDEN_RATIO_PRIME GOLDEN_RATIO_64
+#else
+#error Wordsize not 32 or 64
+#endif
+
+/*
+ * This hash multiplies the input by a large odd number and takes the
+ * high bits.  Since multiplication propagates changes to the most
+ * significant end only, it is essential that the high bits of the
+ * product be used for the hash value.
+ *
+ * Chuck Lever verified the effectiveness of this technique:
+ * http://www.citi.umich.edu/techreports/reports/citi-tr-00-1.pdf
+ *
+ * Although a random odd number will do, it turns out that the golden
+ * ratio phi = (sqrt(5)-1)/2, or its negative, has particularly nice
+ * properties.  (See Knuth vol 3, section 6.4, exercise 9.)
+ *
+ * These are the negative, (1 - phi) = phi**2 = (3 - sqrt(5))/2,
+ * which is very slightly easier to multiply by and makes no
+ * difference to the hash distribution.
+ */
+#define GOLDEN_RATIO_32 0x61C88647
+#define GOLDEN_RATIO_64 0x61C8864680B583EBull
+
+struct hlist_head {
+	struct hlist_node *first;
+};
+
+struct hlist_node {
+	struct hlist_node *next, **pprev;
+};
+
+/*
+ * Architectures might want to move the poison pointer offset
+ * into some well-recognized area such as 0xdead000000000000,
+ * that is also not mappable by user-space exploits:
+ */
+#ifdef CONFIG_ILLEGAL_POINTER_VALUE
+# define POISON_POINTER_DELTA _AC(CONFIG_ILLEGAL_POINTER_VALUE, UL)
+#else
+# define POISON_POINTER_DELTA 0
+#endif
+
+/*
+ * These are non-NULL pointers that will result in page faults
+ * under normal circumstances, used to verify that nobody uses
+ * non-initialized list entries.
+ */
+#define LIST_POISON1 ((void *) 0x00100100 + POISON_POINTER_DELTA)
+#define LIST_POISON2 ((void *) 0x00200200 + POISON_POINTER_DELTA)
+
+/*
+ * Double linked lists with a single pointer list head.
+ * Mostly useful for hash tables where the two pointer list head is
+ * too wasteful.
+ * You lose the ability to access the tail in O(1).
+ */
+
+#define HLIST_HEAD_INIT { .first = NULL }
+#define HLIST_HEAD(name) struct hlist_head name = { .first = NULL }
+#define INIT_HLIST_HEAD(ptr) ((ptr)->first = NULL)
+static inline void INIT_HLIST_NODE(struct hlist_node *h)
+{
+	h->next = NULL;
+	h->pprev = NULL;
+}
+
+static inline int hlist_unhashed(const struct hlist_node *h)
+{
+	return !h->pprev;
+}
+
+static inline int hlist_empty(const struct hlist_head *h)
+{
+	return !h->first;
+}
+
+static inline void __hlist_del(struct hlist_node *n)
+{
+	struct hlist_node *next = n->next;
+	struct hlist_node **pprev = n->pprev;
+
+	*pprev = next;
+	if (next)
+		next->pprev = pprev;
+}
+
+static inline void hlist_del(struct hlist_node *n)
+{
+	__hlist_del(n);
+	n->next = LIST_POISON1;
+	n->pprev = LIST_POISON2;
+}
+
+static inline void hlist_del_init(struct hlist_node *n)
+{
+	if (!hlist_unhashed(n)) {
+		__hlist_del(n);
+		INIT_HLIST_NODE(n);
+	}
+}
+
+static inline void hlist_add_head(struct hlist_node *n, struct hlist_head *h)
+{
+	struct hlist_node *first = h->first;
+
+	n->next = first;
+	if (first)
+		first->pprev = &n->next;
+	h->first = n;
+	n->pprev = &h->first;
+}
+
+/* next must be != NULL */
+static inline void hlist_add_before(struct hlist_node *n,
+					struct hlist_node *next)
+{
+	n->pprev = next->pprev;
+	n->next = next;
+	next->pprev = &n->next;
+	*(n->pprev) = n;
+}
+
+static inline void hlist_add_behind(struct hlist_node *n,
+				    struct hlist_node *prev)
+{
+	n->next = prev->next;
+	prev->next = n;
+	n->pprev = &prev->next;
+
+	if (n->next)
+		n->next->pprev  = &n->next;
+}
+
+/* after that we'll appear to be on some hlist and hlist_del will work */
+static inline void hlist_add_fake(struct hlist_node *n)
+{
+	n->pprev = &n->next;
+}
+
+/*
+ * Move a list from one list head to another. Fixup the pprev
+ * reference of the first entry if it exists.
+ */
+static inline void hlist_move_list(struct hlist_head *old,
+				   struct hlist_head *new)
+{
+	new->first = old->first;
+	if (new->first)
+		new->first->pprev = &new->first;
+	old->first = NULL;
+}
+
+#define hlist_entry(ptr, type, member) container_of(ptr, type, member)
+
+#define hlist_for_each(pos, head) \
+	for (pos = (head)->first; pos; pos = pos->next)
+
+#define hlist_for_each_safe(pos, n, head) \
+	for (pos = (head)->first; pos && ({ n = pos->next; 1; }); \
+	     pos = n)
+
+#define hlist_entry_safe(ptr, type, member) \
+	({ typeof(ptr) ____ptr = (ptr); \
+	   ____ptr ? hlist_entry(____ptr, type, member) : NULL; \
+	})
+
+/**
+ * hlist_for_each_entry	- iterate over list of given type
+ * @pos:the type * to use as a loop cursor.
+ * @head:the head for your list.
+ * @member:the name of the hlist_node within the struct.
+ */
+#define hlist_for_each_entry(pos, head, member)				\
+	for (pos = hlist_entry_safe((head)->first, typeof(*(pos)), member);\
+	     pos;							\
+	     pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member))
+
+/**
+ * hlist_for_each_entry_continue
+ * iterate over a hlist continuing after current point
+ * @pos:the type * to use as a loop cursor.
+ * @member:the name of the hlist_node within the struct.
+ */
+#define hlist_for_each_entry_continue(pos, member)			\
+	for (pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member);\
+	     pos;							\
+	     pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member))
+
+/**
+ * hlist_for_each_entry_from
+ * iterate over a hlist continuing from current point
+ * @pos:	the type * to use as a loop cursor.
+ * @member:	the name of the hlist_node within the struct.
+ */
+#define hlist_for_each_entry_from(pos, member)				\
+	for (; pos;							\
+	     pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member))
+
+/**
+ * hlist_for_each_entry_safe
+ * iterate over list of given type safe against removal of list entry
+ * @pos:the type * to use as a loop cursor.
+ * @n:another &struct hlist_node to use as temporary storage
+ * @head:the head for your list.
+ * @member:the name of the hlist_node within the struct.
+ */
+#define hlist_for_each_entry_safe(pos, n, head, member) 		\
+	for (pos = hlist_entry_safe((head)->first, typeof(*pos), member);\
+		pos && ({ n = pos->member.next; 1; });			\
+		pos = hlist_entry_safe(n, typeof(*pos), member))
+
+static inline u32 __hash_32(u32 val)
+{
+	return val * GOLDEN_RATIO_32;
+}
+
+static inline u32 hash_32(u32 val, unsigned int bits)
+{
+	/* High bits are more random, so use them. */
+	return __hash_32(val) >> (32 - bits);
+}
+
+static __always_inline u32 hash_64(u64 val, unsigned int bits)
+{
+#if BITS_PER_LONG == 64
+	/* 64x64-bit multiply is efficient on all 64-bit processors */
+	return val * GOLDEN_RATIO_64 >> (64 - bits);
+#else
+	/* Hash 64 bits using only 32x32-bit multiply. */
+	return hash_32((u32)val ^ __hash_32(val >> 32), bits);
+#endif
+}
+
+/**
+ * ilog2 - log of base 2 of 32-bit or a 64-bit unsigned value
+ * @n - parameter
+ *
+ * constant-capable log of base 2 calculation
+ * - this can be used to initialise global variables from constant data, hence
+ *   the massive ternary operator construction
+ *
+ * selects the appropriately-sized optimised version depending on sizeof(n)
+ */
+#define ilog2(n)				\
+(								\
+	(n) & (1ULL << 63) ? 63 :	\
+	(n) & (1ULL << 62) ? 62 :	\
+	(n) & (1ULL << 61) ? 61 :	\
+	(n) & (1ULL << 60) ? 60 :	\
+	(n) & (1ULL << 59) ? 59 :	\
+	(n) & (1ULL << 58) ? 58 :	\
+	(n) & (1ULL << 57) ? 57 :	\
+	(n) & (1ULL << 56) ? 56 :	\
+	(n) & (1ULL << 55) ? 55 :	\
+	(n) & (1ULL << 54) ? 54 :	\
+	(n) & (1ULL << 53) ? 53 :	\
+	(n) & (1ULL << 52) ? 52 :	\
+	(n) & (1ULL << 51) ? 51 :	\
+	(n) & (1ULL << 50) ? 50 :	\
+	(n) & (1ULL << 49) ? 49 :	\
+	(n) & (1ULL << 48) ? 48 :	\
+	(n) & (1ULL << 47) ? 47 :	\
+	(n) & (1ULL << 46) ? 46 :	\
+	(n) & (1ULL << 45) ? 45 :	\
+	(n) & (1ULL << 44) ? 44 :	\
+	(n) & (1ULL << 43) ? 43 :	\
+	(n) & (1ULL << 42) ? 42 :	\
+	(n) & (1ULL << 41) ? 41 :	\
+	(n) & (1ULL << 40) ? 40 :	\
+	(n) & (1ULL << 39) ? 39 :	\
+	(n) & (1ULL << 38) ? 38 :	\
+	(n) & (1ULL << 37) ? 37 :	\
+	(n) & (1ULL << 36) ? 36 :	\
+	(n) & (1ULL << 35) ? 35 :	\
+	(n) & (1ULL << 34) ? 34 :	\
+	(n) & (1ULL << 33) ? 33 :	\
+	(n) & (1ULL << 32) ? 32 :	\
+	(n) & (1ULL << 31) ? 31 :	\
+	(n) & (1ULL << 30) ? 30 :	\
+	(n) & (1ULL << 29) ? 29 :	\
+	(n) & (1ULL << 28) ? 28 :	\
+	(n) & (1ULL << 27) ? 27 :	\
+	(n) & (1ULL << 26) ? 26 :	\
+	(n) & (1ULL << 25) ? 25 :	\
+	(n) & (1ULL << 24) ? 24 :	\
+	(n) & (1ULL << 23) ? 23 :	\
+	(n) & (1ULL << 22) ? 22 :	\
+	(n) & (1ULL << 21) ? 21 :	\
+	(n) & (1ULL << 20) ? 20 :	\
+	(n) & (1ULL << 19) ? 19 :	\
+	(n) & (1ULL << 18) ? 18 :	\
+	(n) & (1ULL << 17) ? 17 :	\
+	(n) & (1ULL << 16) ? 16 :	\
+	(n) & (1ULL << 15) ? 15 :	\
+	(n) & (1ULL << 14) ? 14 :	\
+	(n) & (1ULL << 13) ? 13 :	\
+	(n) & (1ULL << 12) ? 12 :	\
+	(n) & (1ULL << 11) ? 11 :	\
+	(n) & (1ULL << 10) ? 10 :	\
+	(n) & (1ULL <<  9) ?  9 :	\
+	(n) & (1ULL <<  8) ?  8 :	\
+	(n) & (1ULL <<  7) ?  7 :	\
+	(n) & (1ULL <<  6) ?  6 :	\
+	(n) & (1ULL <<  5) ?  5 :	\
+	(n) & (1ULL <<  4) ?  4 :	\
+	(n) & (1ULL <<  3) ?  3 :	\
+	(n) & (1ULL <<  2) ?  2 :	\
+	(n) & (1ULL <<  1) ?  1 : 0	\
+)
+
+#define DEFINE_HASHTABLE(name, bits)					\
+	struct hlist_head name[1 << (bits)] =				\
+			{ [0 ... ((1 << (bits)) - 1)] = HLIST_HEAD_INIT }
+
+#define DECLARE_HASHTABLE(name, bits)					\
+	struct hlist_head name[1 << (bits)]
+
+#define HASH_SIZE(name) (ARRAY_SIZE(name))
+#define HASH_BITS(name) ilog2(HASH_SIZE(name))
+
+/* Use hash_32 when possible to allow for fast 32bit hashing in 64bit kernels*/
+#define hash_min(val, bits)						\
+	(sizeof(val) <= 4 ? hash_32(val, bits) : hash_long(val, bits))
+
+static inline void __hash_init(struct hlist_head *ht, unsigned int sz)
+{
+	unsigned int i;
+
+	for (i = 0; i < sz; i++)
+		INIT_HLIST_HEAD(&ht[i]);
+}
+
+/**
+ * hash_init - initialize a hash table
+ * @hashtable: hashtable to be initialized
+ *
+ * Calculates the size of the hashtable from the given parameter, otherwise
+ * same as hash_init_size.
+ *
+ * This has to be a macro since HASH_BITS() will not work on pointers since
+ * it calculates the size during preprocessing.
+ */
+#define hash_init(hashtable) __hash_init(hashtable, HASH_SIZE(hashtable))
+
+/**
+ * hash_add - add an object to a hashtable
+ * @hashtable: hashtable to add to
+ * @node: the &struct hlist_node of the object to be added
+ * @key: the key of the object to be added
+ */
+#define hash_add(hashtable, node, key)					\
+	hlist_add_head(node, &hashtable[hash_min(key, HASH_BITS(hashtable))])
+
+/**
+ * hash_hashed - check whether an object is in any hashtable
+ * @node: the &struct hlist_node of the object to be checked
+ */
+static inline bool hash_hashed(struct hlist_node *node)
+{
+	return !hlist_unhashed(node);
+}
+
+static inline bool __hash_empty(struct hlist_head *ht, unsigned int sz)
+{
+	unsigned int i;
+
+	for (i = 0; i < sz; i++)
+		if (!hlist_empty(&ht[i]))
+			return false;
+
+	return true;
+}
+
+/**
+ * hash_empty - check whether a hashtable is empty
+ * @hashtable: hashtable to check
+ *
+ * This has to be a macro since HASH_BITS() will not work on pointers since
+ * it calculates the size during preprocessing.
+ */
+#define hash_empty(hashtable) __hash_empty(hashtable, HASH_SIZE(hashtable))
+
+/**
+ * hash_del - remove an object from a hashtable
+ * @node: &struct hlist_node of the object to remove
+ */
+static inline void hash_del(struct hlist_node *node)
+{
+	hlist_del_init(node);
+}
+
+/**
+ * hash_for_each - iterate over a hashtable
+ * @name: hashtable to iterate
+ * @bkt: integer to use as bucket loop cursor
+ * @obj: the type * to use as a loop cursor for each entry
+ * @member: the name of the hlist_node within the struct
+ */
+#define hash_for_each(name, bkt, obj, member)				\
+	for ((bkt) = 0, obj = NULL; obj == NULL && (bkt) < HASH_SIZE(name);\
+			(bkt)++)\
+		hlist_for_each_entry(obj, &name[bkt], member)
+
+/**
+ * hash_for_each_safe - iterate over a hashtable safe against removal of
+ * hash entry
+ * @name: hashtable to iterate
+ * @bkt: integer to use as bucket loop cursor
+ * @tmp: a &struct used for temporary storage
+ * @obj: the type * to use as a loop cursor for each entry
+ * @member: the name of the hlist_node within the struct
+ */
+#define hash_for_each_safe(name, bkt, tmp, obj, member)			\
+	for ((bkt) = 0, obj = NULL; obj == NULL && (bkt) < HASH_SIZE(name);\
+			(bkt)++)\
+		hlist_for_each_entry_safe(obj, tmp, &name[bkt], member)
+
+/**
+ * hash_for_each_possible - iterate over all possible objects hashing to the
+ * same bucket
+ * @name: hashtable to iterate
+ * @obj: the type * to use as a loop cursor for each entry
+ * @member: the name of the hlist_node within the struct
+ * @key: the key of the objects to iterate over
+ */
+#define hash_for_each_possible(name, obj, member, key)			\
+	hlist_for_each_entry(obj, &name[hash_min(key, HASH_BITS(name))], member)
+
+#endif
diff --git a/include/erofs/internal.h b/include/erofs/internal.h
index b7ce6f8..33a72b5 100644
--- a/include/erofs/internal.h
+++ b/include/erofs/internal.h
@@ -59,7 +59,7 @@ struct erofs_sb_info {
 extern struct erofs_sb_info sbi;
 
 struct erofs_inode {
-	struct list_head i_hash, i_subdirs;
+	struct list_head i_hash, i_subdirs, i_xattrs;
 
 	unsigned int i_count;
 	struct erofs_inode *i_parent;
diff --git a/include/erofs/xattr.h b/include/erofs/xattr.h
new file mode 100644
index 0000000..5fd6a59
--- /dev/null
+++ b/include/erofs/xattr.h
@@ -0,0 +1,44 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * erofs_utils/include/erofs/xattr.h
+ */
+#ifndef __EROFS_XATTR_H
+#define __EROFS_XATTR_H
+
+#include "internal.h"
+
+#define EROFS_INODE_XATTR_ICOUNT(_size)	({\
+	u32 __size = le16_to_cpu(_size); \
+	((__size) == 0) ? 0 : \
+	(_size - sizeof(struct erofs_xattr_ibody_header)) / \
+	sizeof(struct erofs_xattr_entry) + 1; })
+
+#ifndef XATTR_USER_PREFIX
+#define XATTR_USER_PREFIX	"user."
+#endif
+#ifndef XATTR_USER_PREFIX_LEN
+#define XATTR_USER_PREFIX_LEN (sizeof(XATTR_USER_PREFIX) - 1)
+#endif
+#ifndef XATTR_SECURITY_PREFIX
+#define XATTR_SECURITY_PREFIX	"security."
+#endif
+#ifndef XATTR_SECURITY_PREFIX_LEN
+#define XATTR_SECURITY_PREFIX_LEN (sizeof(XATTR_SECURITY_PREFIX) - 1)
+#endif
+#ifndef XATTR_TRUSTED_PREFIX
+#define XATTR_TRUSTED_PREFIX	"trusted."
+#endif
+#ifndef XATTR_TRUSTED_PREFIX_LEN
+#define XATTR_TRUSTED_PREFIX_LEN (sizeof(XATTR_TRUSTED_PREFIX) - 1)
+#endif
+#ifndef XATTR_NAME_POSIX_ACL_ACCESS
+#define XATTR_NAME_POSIX_ACL_ACCESS "system.posix_acl_access"
+#endif
+#ifndef XATTR_NAME_POSIX_ACL_DEFAULT
+#define XATTR_NAME_POSIX_ACL_DEFAULT "system.posix_acl_default"
+#endif
+
+int erofs_prepare_xattr_ibody(const char *path, struct list_head *ixattrs);
+char *erofs_export_xattr_ibody(struct list_head *ixattrs, unsigned int size);
+
+#endif
diff --git a/lib/Makefile.am b/lib/Makefile.am
index dea82f7..1ff81f9 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -2,7 +2,8 @@
 # Makefile.am
 
 noinst_LTLIBRARIES = liberofs.la
-liberofs_la_SOURCES = config.c io.c cache.c inode.c compress.c compressor.c
+liberofs_la_SOURCES = config.c io.c cache.c inode.c xattr.c \
+		      compress.c compressor.c
 liberofs_la_CFLAGS = -Wall -Werror -I$(top_srcdir)/include
 if ENABLE_LZ4
 liberofs_la_CFLAGS += ${LZ4_CFLAGS}
diff --git a/lib/inode.c b/lib/inode.c
index 581f263..6521a28 100644
--- a/lib/inode.c
+++ b/lib/inode.c
@@ -18,6 +18,7 @@
 #include "erofs/cache.h"
 #include "erofs/io.h"
 #include "erofs/compress.h"
+#include "erofs/xattr.h"
 
 struct erofs_sb_info sbi;
 
@@ -363,9 +364,11 @@ static bool erofs_bh_flush_write_inode(struct erofs_buffer_head *bh)
 
 	/* let's support v1 currently */
 	struct erofs_inode_v1 v1 = {0};
+	const u16 icount = EROFS_INODE_XATTR_ICOUNT(inode->xattr_isize);
 	int ret;
 
 	v1.i_advise = cpu_to_le16(0 | (inode->data_mapping_mode << 1));
+	v1.i_xattr_icount = cpu_to_le16(icount);
 	v1.i_mode = cpu_to_le16(inode->i_mode);
 	v1.i_nlink = cpu_to_le16(inode->i_nlink);
 	v1.i_size = cpu_to_le32((u32)inode->i_size);
@@ -398,6 +401,20 @@ static bool erofs_bh_flush_write_inode(struct erofs_buffer_head *bh)
 		return false;
 	off += inode->inode_isize;
 
+	if (inode->xattr_isize) {
+		char *xattrs = erofs_export_xattr_ibody(&inode->i_xattrs,
+							inode->xattr_isize);
+		if (IS_ERR(xattrs))
+			return false;
+
+		ret = dev_write(xattrs, off, inode->xattr_isize);
+		free(xattrs);
+		if (ret)
+			return false;
+
+		off += inode->xattr_isize;
+	}
+
 	if (inode->extent_isize) {
 		/* write compression metadata */
 		off = Z_EROFS_VLE_EXTENT_ALIGN(off);
@@ -612,6 +629,7 @@ struct erofs_inode *erofs_new_inode(void)
 	inode->i_count = 1;
 
 	init_list_head(&inode->i_subdirs);
+	init_list_head(&inode->i_xattrs);
 	inode->xattr_isize = 0;
 	inode->extent_isize = 0;
 
@@ -699,6 +717,11 @@ struct erofs_inode *erofs_mkfs_build_tree(struct erofs_inode *dir)
 	struct dirent *dp;
 	struct erofs_dentry *d;
 
+	ret = erofs_prepare_xattr_ibody(dir->i_srcpath, &dir->i_xattrs);
+	if (ret < 0)
+		return ERR_PTR(ret);
+	dir->xattr_isize = ret;
+
 	if (!S_ISDIR(dir->i_mode)) {
 		if (S_ISLNK(dir->i_mode)) {
 			char *const symlink = malloc(dir->i_size);
diff --git a/lib/xattr.c b/lib/xattr.c
new file mode 100644
index 0000000..1ab959b
--- /dev/null
+++ b/lib/xattr.c
@@ -0,0 +1,289 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * erofs_utils/lib/xattr.c
+ */
+#include <stdlib.h>
+#include <sys/xattr.h>
+#ifdef HAVE_LINUX_XATTR_H
+#include <linux/xattr.h>
+#endif
+#include "erofs/print.h"
+#include "erofs/hashtable.h"
+#include "erofs/xattr.h"
+
+#define EA_HASHTABLE_BITS 16
+
+struct xattr_item {
+	const char *kvbuf;
+	unsigned int hash[2], len[2], count;
+	u8 prefix;
+	struct hlist_node node;
+};
+
+struct inode_xattr_node {
+	struct list_head list;
+	struct xattr_item *item;
+};
+
+static DECLARE_HASHTABLE(ea_hashtable, EA_HASHTABLE_BITS);
+
+static struct xattr_prefix {
+	const char *prefix;
+	u16 prefix_len;
+} prefix[] = {
+	[EROFS_XATTR_INDEX_USER] = {
+		XATTR_USER_PREFIX,
+		XATTR_USER_PREFIX_LEN
+	}, [EROFS_XATTR_INDEX_POSIX_ACL_ACCESS] = {
+		XATTR_NAME_POSIX_ACL_ACCESS,
+		sizeof(XATTR_NAME_POSIX_ACL_ACCESS) - 1
+	}, [EROFS_XATTR_INDEX_POSIX_ACL_DEFAULT] = {
+		XATTR_NAME_POSIX_ACL_DEFAULT,
+		sizeof(XATTR_NAME_POSIX_ACL_DEFAULT) - 1
+	}, [EROFS_XATTR_INDEX_TRUSTED] = {
+		XATTR_TRUSTED_PREFIX,
+		XATTR_TRUSTED_PREFIX_LEN
+	}, [EROFS_XATTR_INDEX_SECURITY] = {
+		XATTR_SECURITY_PREFIX,
+		XATTR_SECURITY_PREFIX_LEN
+	}
+};
+
+static unsigned int BKDRHash(char *str, unsigned int len)
+{
+	const unsigned int seed = 131313;
+	unsigned int hash = 0;
+
+	while (len) {
+		hash = hash * seed + (*str++);
+		--len;
+	}
+	return hash;
+}
+
+static unsigned int xattr_item_hash(u8 prefix, char *buf,
+				    unsigned int len[2], unsigned int hash[2])
+{
+	hash[0] = BKDRHash(buf, len[0]);	/* key */
+	hash[1] = BKDRHash(buf + len[0], len[1]);	/* value */
+
+	return prefix ^ hash[0] ^ hash[1];
+}
+
+static struct xattr_item *xattr_item_get(u8 prefix, char *kvbuf,
+					 unsigned int len[2])
+{
+	struct xattr_item *item;
+	unsigned int hash[2], hkey;
+
+	hkey = xattr_item_hash(prefix, kvbuf, len, hash);
+
+	hash_for_each_possible(ea_hashtable, item, node, hkey) {
+		if (prefix == item->prefix &&
+		    item->len[0] == len[0] && item->len[1] == len[1] &&
+		    item->hash[0] == hash[0] && item->hash[1] == hash[1] &&
+		    !memcmp(kvbuf, item->kvbuf, len[0] + len[1])) {
+			free(kvbuf);
+			++item->count;
+			return item;
+		}
+	}
+
+	item = malloc(sizeof(*item));
+	if (!item) {
+		free(kvbuf);
+		return ERR_PTR(-ENOMEM);
+	}
+	INIT_HLIST_NODE(&item->node);
+	item->count = 1;
+	item->kvbuf = kvbuf;
+	item->len[0] = len[0];
+	item->len[1] = len[1];
+	item->hash[0] = hash[0];
+	item->hash[1] = hash[1];
+	item->prefix = prefix;
+	hash_add(ea_hashtable, &item->node, hkey);
+	return item;
+}
+
+static bool match_prefix(const char *key, u8 *index, u16 *len)
+{
+	struct xattr_prefix *p;
+
+	for (p = prefix; p < prefix + ARRAY_SIZE(prefix); ++p) {
+		if (p->prefix && !strncmp(p->prefix, key, p->prefix_len)) {
+			*len = p->prefix_len;
+			*index = p - prefix;
+			return true;
+		}
+	}
+	return false;
+}
+
+static struct xattr_item *parse_one_xattr(const char *path, const char *key,
+					  unsigned int keylen)
+{
+	ssize_t ret;
+	u8 prefix;
+	u16 prefixlen;
+	unsigned int len[2];
+	char *kvbuf;
+
+	erofs_dbg("parse xattr [%s] of %s", path, key);
+
+	if (!match_prefix(key, &prefix, &prefixlen))
+		return ERR_PTR(-ENODATA);
+
+	DBG_BUGON(keylen < prefixlen);
+
+	/* determine length of the value */
+	ret = lgetxattr(path, key, NULL, 0);
+	if (ret < 0)
+		return ERR_PTR(-errno);
+	len[1] = ret;
+
+	/* allocate key-value buffer */
+	len[0] = keylen - prefixlen;
+
+	kvbuf = malloc(len[0] + len[1]);
+	if (!kvbuf)
+		return ERR_PTR(-ENOMEM);
+
+	memcpy(kvbuf, key + prefixlen, len[0]);
+	if (len[1]) {
+		/* copy value to buffer */
+		ret = lgetxattr(path, key, kvbuf + len[0], len[1]);
+		if (ret < 0) {
+			free(kvbuf);
+			return ERR_PTR(-errno);
+		}
+		if (len[1] != ret) {
+			erofs_err("xattr value changed just now (%u-> %ld)",
+				  len[1], ret);
+			len[1] = ret;
+		}
+	}
+	return xattr_item_get(prefix, kvbuf, len);
+}
+
+static int inode_xattr_add(struct list_head *hlist, struct xattr_item *item)
+{
+	struct inode_xattr_node *node = malloc(sizeof(*node));
+
+	if (!node)
+		return -ENOMEM;
+	init_list_head(&node->list);
+	node->item = item;
+	list_add(&node->list, hlist);
+	return 0;
+}
+
+static int read_xattrs_from_file(const char *path, struct list_head *ixattrs)
+{
+	int ret = 0;
+	char *keylst, *key;
+	ssize_t kllen = llistxattr(path, NULL, 0);
+
+	if (kllen < 0 && errno != ENODATA)
+		return -errno;
+	if (kllen <= 1)
+		return 0;
+
+	keylst = malloc(kllen);
+	if (!keylst)
+		return -errno;
+
+	/* copy the list of attribute keys to the buffer.*/
+	kllen = llistxattr(path, keylst, kllen);
+	if (kllen < 0) {
+		ret = -errno;
+		goto err;
+	}
+
+	/*
+	 * loop over the list of zero terminated strings with the
+	 * attribute keys. Use the remaining buffer length to determine
+	 * the end of the list.
+	 */
+	key = keylst;
+	while (kllen > 0) {
+		unsigned int keylen = strlen(key);
+		struct xattr_item *item = parse_one_xattr(path, key, keylen);
+
+		if (IS_ERR(item)) {
+			ret = PTR_ERR(item);
+			goto err;
+		}
+
+		if (ixattrs) {
+			ret = inode_xattr_add(ixattrs, item);
+			if (ret < 0)
+				goto err;
+		}
+		kllen -= keylen + 1;
+		key += keylen + 1;
+	}
+err:
+	free(keylst);
+	return ret;
+
+}
+
+int erofs_prepare_xattr_ibody(const char *path, struct list_head *ixattrs)
+{
+	int ret;
+	struct inode_xattr_node *node;
+
+	ret = read_xattrs_from_file(path, ixattrs);
+	if (ret < 0)
+		return ret;
+
+	if (list_empty(ixattrs))
+		return 0;
+
+	/* get xattr ibody size */
+	ret = sizeof(struct erofs_xattr_ibody_header);
+	list_for_each_entry(node, ixattrs, list) {
+		const struct xattr_item *item = node->item;
+
+		ret += sizeof(struct erofs_xattr_entry);
+		ret = EROFS_XATTR_ALIGN(ret + item->len[0] + item->len[1]);
+	}
+	return ret;
+}
+
+char *erofs_export_xattr_ibody(struct list_head *ixattrs, unsigned int size)
+{
+	char *buf;
+	struct inode_xattr_node *node;
+	struct erofs_xattr_ibody_header *header;
+	unsigned int p;
+
+	buf = calloc(1, size);
+	if (!buf)
+		return ERR_PTR(-ENOMEM);
+
+	header = (struct erofs_xattr_ibody_header *)buf;
+	header->h_shared_count = 0;
+
+	p = sizeof(struct erofs_xattr_ibody_header);
+	list_for_each_entry(node, ixattrs, list) {
+		struct xattr_item *const item = node->item;
+		const struct erofs_xattr_entry entry = {
+			.e_name_index = item->prefix,
+			.e_name_len = item->len[0],
+			.e_value_size = cpu_to_le16(item->len[1])
+		};
+
+		memcpy(buf + p, &entry, sizeof(entry));
+		p += sizeof(struct erofs_xattr_entry);
+		memcpy(buf + p, item->kvbuf, item->len[0] + item->len[1]);
+		p = EROFS_XATTR_ALIGN(p + item->len[0] + item->len[1]);
+
+		list_del(&node->list);
+		free(node);
+	}
+	DBG_BUGON(p > size);
+	return buf;
+}
+
-- 
2.17.1

^ permalink raw reply related	[flat|nested] 12+ messages in thread

* [PATCH v3 1/2] erofs-utils: introduce inline xattr support
  2019-08-11 17:10     ` [PATCH v2] erofs-utils: introduce preliminary " Gao Xiang
@ 2019-10-05 14:20       ` Gao Xiang via Linux-erofs
  2019-10-05 14:20         ` [PATCH 2/2] erofs-utils: introduce shared " Gao Xiang via Linux-erofs
  2019-10-05 16:44         ` [PATCH v3 1/2] erofs-utils: introduce inline " Li Guifu
  0 siblings, 2 replies; 12+ messages in thread
From: Gao Xiang via Linux-erofs @ 2019-10-05 14:20 UTC (permalink / raw)
  To: Li Guifu, Chao Yu, linux-erofs; +Cc: htyuxe+dhbrei4sq0df8

From: "htyuxe+dhbrei4sq0df8@grr.la" <htyuxe+dhbrei4sq0df8@grr.la>

Load xattrs from source files and pack them into target image.

Signed-off-by: htyuxe+dhbrei4sq0df8@grr.la <htyuxe+dhbrei4sq0df8@grr.la>
Signed-off-by: Li Guifu <blucerlee@gmail.com>
Signed-off-by: Gao Xiang <hsiangkao@aol.com>
---

This is a cleanup xattr implementation for new erofs-utils
compared with old internal commerical mess, which will be
included in erofs-utils v1.0 eventually.

v3:
 - adapt shared xattr patch;
 - fix alignment after xattr applied.

 configure.ac              |   1 +
 include/erofs/defs.h      |   6 +
 include/erofs/hashtable.h | 462 ++++++++++++++++++++++++++++++++++++++
 include/erofs/internal.h  |   2 +-
 include/erofs/xattr.h     |  48 ++++
 lib/Makefile.am           |   3 +-
 lib/inode.c               |  31 ++-
 lib/xattr.c               | 296 ++++++++++++++++++++++++
 8 files changed, 845 insertions(+), 4 deletions(-)
 create mode 100644 include/erofs/hashtable.h
 create mode 100644 include/erofs/xattr.h
 create mode 100644 lib/xattr.c

diff --git a/configure.ac b/configure.ac
index 07e034e..8c31cb7 100644
--- a/configure.ac
+++ b/configure.ac
@@ -75,6 +75,7 @@ AC_CHECK_HEADERS(m4_flatten([
 	linux/falloc.h
 	linux/fs.h
 	linux/types.h
+	linux/xattr.h
 	limits.h
 	stddef.h
 	stdint.h
diff --git a/include/erofs/defs.h b/include/erofs/defs.h
index 15db4e3..c4cdb2d 100644
--- a/include/erofs/defs.h
+++ b/include/erofs/defs.h
@@ -151,6 +151,12 @@ typedef int64_t         s64;
 #define BITS_PER_BYTE       8
 #define BITS_TO_LONGS(nr)   DIV_ROUND_UP(nr, BITS_PER_BYTE * sizeof(long))
 
+#ifdef __SIZEOF_LONG__
+#define BITS_PER_LONG (__CHAR_BIT__ * __SIZEOF_LONG__)
+#else
+#define BITS_PER_LONG __WORDSIZE
+#endif
+
 #define BUG_ON(cond)        assert(!(cond))
 
 #ifdef NDEBUG
diff --git a/include/erofs/hashtable.h b/include/erofs/hashtable.h
new file mode 100644
index 0000000..ab57b56
--- /dev/null
+++ b/include/erofs/hashtable.h
@@ -0,0 +1,462 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * erofs_utils/include/erofs/hashtable.h
+ *
+ * Original code taken from 'linux/include/linux/hash{,table}.h'
+ */
+#ifndef __EROFS_HASHTABLE_H
+#define __EROFS_HASHTABLE_H
+
+/*
+ * Fast hashing routine for ints,  longs and pointers.
+ * (C) 2002 Nadia Yvette Chambers, IBM
+ */
+
+/*
+ * Statically sized hash table implementation
+ * (C) 2012  Sasha Levin <levinsasha928@gmail.com>
+ */
+
+#include "defs.h"
+
+/*
+ * The "GOLDEN_RATIO_PRIME" is used in ifs/btrfs/brtfs_inode.h and
+ * fs/inode.c.  It's not actually prime any more (the previous primes
+ * were actively bad for hashing), but the name remains.
+ */
+#if BITS_PER_LONG == 32
+#define GOLDEN_RATIO_PRIME GOLDEN_RATIO_32
+#define hash_long(val, bits) hash_32(val, bits)
+#elif BITS_PER_LONG == 64
+#define hash_long(val, bits) hash_64(val, bits)
+#define GOLDEN_RATIO_PRIME GOLDEN_RATIO_64
+#else
+#error Wordsize not 32 or 64
+#endif
+
+/*
+ * This hash multiplies the input by a large odd number and takes the
+ * high bits.  Since multiplication propagates changes to the most
+ * significant end only, it is essential that the high bits of the
+ * product be used for the hash value.
+ *
+ * Chuck Lever verified the effectiveness of this technique:
+ * http://www.citi.umich.edu/techreports/reports/citi-tr-00-1.pdf
+ *
+ * Although a random odd number will do, it turns out that the golden
+ * ratio phi = (sqrt(5)-1)/2, or its negative, has particularly nice
+ * properties.  (See Knuth vol 3, section 6.4, exercise 9.)
+ *
+ * These are the negative, (1 - phi) = phi**2 = (3 - sqrt(5))/2,
+ * which is very slightly easier to multiply by and makes no
+ * difference to the hash distribution.
+ */
+#define GOLDEN_RATIO_32 0x61C88647
+#define GOLDEN_RATIO_64 0x61C8864680B583EBull
+
+struct hlist_head {
+	struct hlist_node *first;
+};
+
+struct hlist_node {
+	struct hlist_node *next, **pprev;
+};
+
+/*
+ * Architectures might want to move the poison pointer offset
+ * into some well-recognized area such as 0xdead000000000000,
+ * that is also not mappable by user-space exploits:
+ */
+#ifdef CONFIG_ILLEGAL_POINTER_VALUE
+# define POISON_POINTER_DELTA _AC(CONFIG_ILLEGAL_POINTER_VALUE, UL)
+#else
+# define POISON_POINTER_DELTA 0
+#endif
+
+/*
+ * These are non-NULL pointers that will result in page faults
+ * under normal circumstances, used to verify that nobody uses
+ * non-initialized list entries.
+ */
+#define LIST_POISON1 ((void *) 0x00100100 + POISON_POINTER_DELTA)
+#define LIST_POISON2 ((void *) 0x00200200 + POISON_POINTER_DELTA)
+
+/*
+ * Double linked lists with a single pointer list head.
+ * Mostly useful for hash tables where the two pointer list head is
+ * too wasteful.
+ * You lose the ability to access the tail in O(1).
+ */
+
+#define HLIST_HEAD_INIT { .first = NULL }
+#define HLIST_HEAD(name) struct hlist_head name = { .first = NULL }
+#define INIT_HLIST_HEAD(ptr) ((ptr)->first = NULL)
+static inline void INIT_HLIST_NODE(struct hlist_node *h)
+{
+	h->next = NULL;
+	h->pprev = NULL;
+}
+
+static inline int hlist_unhashed(const struct hlist_node *h)
+{
+	return !h->pprev;
+}
+
+static inline int hlist_empty(const struct hlist_head *h)
+{
+	return !h->first;
+}
+
+static inline void __hlist_del(struct hlist_node *n)
+{
+	struct hlist_node *next = n->next;
+	struct hlist_node **pprev = n->pprev;
+
+	*pprev = next;
+	if (next)
+		next->pprev = pprev;
+}
+
+static inline void hlist_del(struct hlist_node *n)
+{
+	__hlist_del(n);
+	n->next = LIST_POISON1;
+	n->pprev = LIST_POISON2;
+}
+
+static inline void hlist_del_init(struct hlist_node *n)
+{
+	if (!hlist_unhashed(n)) {
+		__hlist_del(n);
+		INIT_HLIST_NODE(n);
+	}
+}
+
+static inline void hlist_add_head(struct hlist_node *n, struct hlist_head *h)
+{
+	struct hlist_node *first = h->first;
+
+	n->next = first;
+	if (first)
+		first->pprev = &n->next;
+	h->first = n;
+	n->pprev = &h->first;
+}
+
+/* next must be != NULL */
+static inline void hlist_add_before(struct hlist_node *n,
+					struct hlist_node *next)
+{
+	n->pprev = next->pprev;
+	n->next = next;
+	next->pprev = &n->next;
+	*(n->pprev) = n;
+}
+
+static inline void hlist_add_behind(struct hlist_node *n,
+				    struct hlist_node *prev)
+{
+	n->next = prev->next;
+	prev->next = n;
+	n->pprev = &prev->next;
+
+	if (n->next)
+		n->next->pprev  = &n->next;
+}
+
+/* after that we'll appear to be on some hlist and hlist_del will work */
+static inline void hlist_add_fake(struct hlist_node *n)
+{
+	n->pprev = &n->next;
+}
+
+/*
+ * Move a list from one list head to another. Fixup the pprev
+ * reference of the first entry if it exists.
+ */
+static inline void hlist_move_list(struct hlist_head *old,
+				   struct hlist_head *new)
+{
+	new->first = old->first;
+	if (new->first)
+		new->first->pprev = &new->first;
+	old->first = NULL;
+}
+
+#define hlist_entry(ptr, type, member) container_of(ptr, type, member)
+
+#define hlist_for_each(pos, head) \
+	for (pos = (head)->first; pos; pos = pos->next)
+
+#define hlist_for_each_safe(pos, n, head) \
+	for (pos = (head)->first; pos && ({ n = pos->next; 1; }); \
+	     pos = n)
+
+#define hlist_entry_safe(ptr, type, member) \
+	({ typeof(ptr) ____ptr = (ptr); \
+	   ____ptr ? hlist_entry(____ptr, type, member) : NULL; \
+	})
+
+/**
+ * hlist_for_each_entry	- iterate over list of given type
+ * @pos:the type * to use as a loop cursor.
+ * @head:the head for your list.
+ * @member:the name of the hlist_node within the struct.
+ */
+#define hlist_for_each_entry(pos, head, member)				\
+	for (pos = hlist_entry_safe((head)->first, typeof(*(pos)), member);\
+	     pos;							\
+	     pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member))
+
+/**
+ * hlist_for_each_entry_continue
+ * iterate over a hlist continuing after current point
+ * @pos:the type * to use as a loop cursor.
+ * @member:the name of the hlist_node within the struct.
+ */
+#define hlist_for_each_entry_continue(pos, member)			\
+	for (pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member);\
+	     pos;							\
+	     pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member))
+
+/**
+ * hlist_for_each_entry_from
+ * iterate over a hlist continuing from current point
+ * @pos:	the type * to use as a loop cursor.
+ * @member:	the name of the hlist_node within the struct.
+ */
+#define hlist_for_each_entry_from(pos, member)				\
+	for (; pos;							\
+	     pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member))
+
+/**
+ * hlist_for_each_entry_safe
+ * iterate over list of given type safe against removal of list entry
+ * @pos:the type * to use as a loop cursor.
+ * @n:another &struct hlist_node to use as temporary storage
+ * @head:the head for your list.
+ * @member:the name of the hlist_node within the struct.
+ */
+#define hlist_for_each_entry_safe(pos, n, head, member)			\
+	for (pos = hlist_entry_safe((head)->first, typeof(*pos), member);\
+		pos && ({ n = pos->member.next; 1; });			\
+		pos = hlist_entry_safe(n, typeof(*pos), member))
+
+static inline u32 __hash_32(u32 val)
+{
+	return val * GOLDEN_RATIO_32;
+}
+
+static inline u32 hash_32(u32 val, unsigned int bits)
+{
+	/* High bits are more random, so use them. */
+	return __hash_32(val) >> (32 - bits);
+}
+
+static __always_inline u32 hash_64(u64 val, unsigned int bits)
+{
+#if BITS_PER_LONG == 64
+	/* 64x64-bit multiply is efficient on all 64-bit processors */
+	return val * GOLDEN_RATIO_64 >> (64 - bits);
+#else
+	/* Hash 64 bits using only 32x32-bit multiply. */
+	return hash_32((u32)val ^ __hash_32(val >> 32), bits);
+#endif
+}
+
+/**
+ * ilog2 - log of base 2 of 32-bit or a 64-bit unsigned value
+ * @n - parameter
+ *
+ * constant-capable log of base 2 calculation
+ * - this can be used to initialise global variables from constant data, hence
+ *   the massive ternary operator construction
+ *
+ * selects the appropriately-sized optimised version depending on sizeof(n)
+ */
+#define ilog2(n)				\
+(								\
+	(n) & (1ULL << 63) ? 63 :	\
+	(n) & (1ULL << 62) ? 62 :	\
+	(n) & (1ULL << 61) ? 61 :	\
+	(n) & (1ULL << 60) ? 60 :	\
+	(n) & (1ULL << 59) ? 59 :	\
+	(n) & (1ULL << 58) ? 58 :	\
+	(n) & (1ULL << 57) ? 57 :	\
+	(n) & (1ULL << 56) ? 56 :	\
+	(n) & (1ULL << 55) ? 55 :	\
+	(n) & (1ULL << 54) ? 54 :	\
+	(n) & (1ULL << 53) ? 53 :	\
+	(n) & (1ULL << 52) ? 52 :	\
+	(n) & (1ULL << 51) ? 51 :	\
+	(n) & (1ULL << 50) ? 50 :	\
+	(n) & (1ULL << 49) ? 49 :	\
+	(n) & (1ULL << 48) ? 48 :	\
+	(n) & (1ULL << 47) ? 47 :	\
+	(n) & (1ULL << 46) ? 46 :	\
+	(n) & (1ULL << 45) ? 45 :	\
+	(n) & (1ULL << 44) ? 44 :	\
+	(n) & (1ULL << 43) ? 43 :	\
+	(n) & (1ULL << 42) ? 42 :	\
+	(n) & (1ULL << 41) ? 41 :	\
+	(n) & (1ULL << 40) ? 40 :	\
+	(n) & (1ULL << 39) ? 39 :	\
+	(n) & (1ULL << 38) ? 38 :	\
+	(n) & (1ULL << 37) ? 37 :	\
+	(n) & (1ULL << 36) ? 36 :	\
+	(n) & (1ULL << 35) ? 35 :	\
+	(n) & (1ULL << 34) ? 34 :	\
+	(n) & (1ULL << 33) ? 33 :	\
+	(n) & (1ULL << 32) ? 32 :	\
+	(n) & (1ULL << 31) ? 31 :	\
+	(n) & (1ULL << 30) ? 30 :	\
+	(n) & (1ULL << 29) ? 29 :	\
+	(n) & (1ULL << 28) ? 28 :	\
+	(n) & (1ULL << 27) ? 27 :	\
+	(n) & (1ULL << 26) ? 26 :	\
+	(n) & (1ULL << 25) ? 25 :	\
+	(n) & (1ULL << 24) ? 24 :	\
+	(n) & (1ULL << 23) ? 23 :	\
+	(n) & (1ULL << 22) ? 22 :	\
+	(n) & (1ULL << 21) ? 21 :	\
+	(n) & (1ULL << 20) ? 20 :	\
+	(n) & (1ULL << 19) ? 19 :	\
+	(n) & (1ULL << 18) ? 18 :	\
+	(n) & (1ULL << 17) ? 17 :	\
+	(n) & (1ULL << 16) ? 16 :	\
+	(n) & (1ULL << 15) ? 15 :	\
+	(n) & (1ULL << 14) ? 14 :	\
+	(n) & (1ULL << 13) ? 13 :	\
+	(n) & (1ULL << 12) ? 12 :	\
+	(n) & (1ULL << 11) ? 11 :	\
+	(n) & (1ULL << 10) ? 10 :	\
+	(n) & (1ULL <<  9) ?  9 :	\
+	(n) & (1ULL <<  8) ?  8 :	\
+	(n) & (1ULL <<  7) ?  7 :	\
+	(n) & (1ULL <<  6) ?  6 :	\
+	(n) & (1ULL <<  5) ?  5 :	\
+	(n) & (1ULL <<  4) ?  4 :	\
+	(n) & (1ULL <<  3) ?  3 :	\
+	(n) & (1ULL <<  2) ?  2 :	\
+	(n) & (1ULL <<  1) ?  1 : 0	\
+)
+
+#define DEFINE_HASHTABLE(name, bits)					\
+	struct hlist_head name[1 << (bits)] =				\
+			{ [0 ... ((1 << (bits)) - 1)] = HLIST_HEAD_INIT }
+
+#define DECLARE_HASHTABLE(name, bits)					\
+	struct hlist_head name[1 << (bits)]
+
+#define HASH_SIZE(name) (ARRAY_SIZE(name))
+#define HASH_BITS(name) ilog2(HASH_SIZE(name))
+
+/* Use hash_32 when possible to allow for fast 32bit hashing in 64bit kernels*/
+#define hash_min(val, bits)						\
+	(sizeof(val) <= 4 ? hash_32(val, bits) : hash_long(val, bits))
+
+static inline void __hash_init(struct hlist_head *ht, unsigned int sz)
+{
+	unsigned int i;
+
+	for (i = 0; i < sz; i++)
+		INIT_HLIST_HEAD(&ht[i]);
+}
+
+/**
+ * hash_init - initialize a hash table
+ * @hashtable: hashtable to be initialized
+ *
+ * Calculates the size of the hashtable from the given parameter, otherwise
+ * same as hash_init_size.
+ *
+ * This has to be a macro since HASH_BITS() will not work on pointers since
+ * it calculates the size during preprocessing.
+ */
+#define hash_init(hashtable) __hash_init(hashtable, HASH_SIZE(hashtable))
+
+/**
+ * hash_add - add an object to a hashtable
+ * @hashtable: hashtable to add to
+ * @node: the &struct hlist_node of the object to be added
+ * @key: the key of the object to be added
+ */
+#define hash_add(hashtable, node, key)					\
+	hlist_add_head(node, &hashtable[hash_min(key, HASH_BITS(hashtable))])
+
+/**
+ * hash_hashed - check whether an object is in any hashtable
+ * @node: the &struct hlist_node of the object to be checked
+ */
+static inline bool hash_hashed(struct hlist_node *node)
+{
+	return !hlist_unhashed(node);
+}
+
+static inline bool __hash_empty(struct hlist_head *ht, unsigned int sz)
+{
+	unsigned int i;
+
+	for (i = 0; i < sz; i++)
+		if (!hlist_empty(&ht[i]))
+			return false;
+
+	return true;
+}
+
+/**
+ * hash_empty - check whether a hashtable is empty
+ * @hashtable: hashtable to check
+ *
+ * This has to be a macro since HASH_BITS() will not work on pointers since
+ * it calculates the size during preprocessing.
+ */
+#define hash_empty(hashtable) __hash_empty(hashtable, HASH_SIZE(hashtable))
+
+/**
+ * hash_del - remove an object from a hashtable
+ * @node: &struct hlist_node of the object to remove
+ */
+static inline void hash_del(struct hlist_node *node)
+{
+	hlist_del_init(node);
+}
+
+/**
+ * hash_for_each - iterate over a hashtable
+ * @name: hashtable to iterate
+ * @bkt: integer to use as bucket loop cursor
+ * @obj: the type * to use as a loop cursor for each entry
+ * @member: the name of the hlist_node within the struct
+ */
+#define hash_for_each(name, bkt, obj, member)				\
+	for ((bkt) = 0, obj = NULL; obj == NULL && (bkt) < HASH_SIZE(name);\
+			(bkt)++)\
+		hlist_for_each_entry(obj, &name[bkt], member)
+
+/**
+ * hash_for_each_safe - iterate over a hashtable safe against removal of
+ * hash entry
+ * @name: hashtable to iterate
+ * @bkt: integer to use as bucket loop cursor
+ * @tmp: a &struct used for temporary storage
+ * @obj: the type * to use as a loop cursor for each entry
+ * @member: the name of the hlist_node within the struct
+ */
+#define hash_for_each_safe(name, bkt, tmp, obj, member)			\
+	for ((bkt) = 0, obj = NULL; obj == NULL && (bkt) < HASH_SIZE(name);\
+			(bkt)++)\
+		hlist_for_each_entry_safe(obj, tmp, &name[bkt], member)
+
+/**
+ * hash_for_each_possible - iterate over all possible objects hashing to the
+ * same bucket
+ * @name: hashtable to iterate
+ * @obj: the type * to use as a loop cursor for each entry
+ * @member: the name of the hlist_node within the struct
+ * @key: the key of the objects to iterate over
+ */
+#define hash_for_each_possible(name, obj, member, key)			\
+	hlist_for_each_entry(obj, &name[hash_min(key, HASH_BITS(name))], member)
+
+#endif
diff --git a/include/erofs/internal.h b/include/erofs/internal.h
index 5384946..25ce7b5 100644
--- a/include/erofs/internal.h
+++ b/include/erofs/internal.h
@@ -61,7 +61,7 @@ struct erofs_sb_info {
 extern struct erofs_sb_info sbi;
 
 struct erofs_inode {
-	struct list_head i_hash, i_subdirs;
+	struct list_head i_hash, i_subdirs, i_xattrs;
 
 	unsigned int i_count;
 	struct erofs_inode *i_parent;
diff --git a/include/erofs/xattr.h b/include/erofs/xattr.h
new file mode 100644
index 0000000..29df025
--- /dev/null
+++ b/include/erofs/xattr.h
@@ -0,0 +1,48 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * erofs_utils/include/erofs/xattr.h
+ *
+ * Originally contributed by an anonymous person,
+ * heavily changed by Li Guifu <blucerlee@gmail.com>
+ *                and Gao Xiang <xiang@kernel.org>
+ */
+#ifndef __EROFS_XATTR_H
+#define __EROFS_XATTR_H
+
+#include "internal.h"
+
+#define EROFS_INODE_XATTR_ICOUNT(_size)	({\
+	u32 __size = le16_to_cpu(_size); \
+	((__size) == 0) ? 0 : \
+	(_size - sizeof(struct erofs_xattr_ibody_header)) / \
+	sizeof(struct erofs_xattr_entry) + 1; })
+
+#ifndef XATTR_USER_PREFIX
+#define XATTR_USER_PREFIX	"user."
+#endif
+#ifndef XATTR_USER_PREFIX_LEN
+#define XATTR_USER_PREFIX_LEN (sizeof(XATTR_USER_PREFIX) - 1)
+#endif
+#ifndef XATTR_SECURITY_PREFIX
+#define XATTR_SECURITY_PREFIX	"security."
+#endif
+#ifndef XATTR_SECURITY_PREFIX_LEN
+#define XATTR_SECURITY_PREFIX_LEN (sizeof(XATTR_SECURITY_PREFIX) - 1)
+#endif
+#ifndef XATTR_TRUSTED_PREFIX
+#define XATTR_TRUSTED_PREFIX	"trusted."
+#endif
+#ifndef XATTR_TRUSTED_PREFIX_LEN
+#define XATTR_TRUSTED_PREFIX_LEN (sizeof(XATTR_TRUSTED_PREFIX) - 1)
+#endif
+#ifndef XATTR_NAME_POSIX_ACL_ACCESS
+#define XATTR_NAME_POSIX_ACL_ACCESS "system.posix_acl_access"
+#endif
+#ifndef XATTR_NAME_POSIX_ACL_DEFAULT
+#define XATTR_NAME_POSIX_ACL_DEFAULT "system.posix_acl_default"
+#endif
+
+int erofs_prepare_xattr_ibody(const char *path, struct list_head *ixattrs);
+char *erofs_export_xattr_ibody(struct list_head *ixattrs, unsigned int size);
+
+#endif
diff --git a/lib/Makefile.am b/lib/Makefile.am
index dea82f7..1ff81f9 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -2,7 +2,8 @@
 # Makefile.am
 
 noinst_LTLIBRARIES = liberofs.la
-liberofs_la_SOURCES = config.c io.c cache.c inode.c compress.c compressor.c
+liberofs_la_SOURCES = config.c io.c cache.c inode.c xattr.c \
+		      compress.c compressor.c
 liberofs_la_CFLAGS = -Wall -Werror -I$(top_srcdir)/include
 if ENABLE_LZ4
 liberofs_la_CFLAGS += ${LZ4_CFLAGS}
diff --git a/lib/inode.c b/lib/inode.c
index d1f294a..a23bf8f 100644
--- a/lib/inode.c
+++ b/lib/inode.c
@@ -20,6 +20,7 @@
 #include "erofs/cache.h"
 #include "erofs/io.h"
 #include "erofs/compress.h"
+#include "erofs/xattr.h"
 
 struct erofs_sb_info sbi;
 
@@ -365,6 +366,7 @@ fail:
 static bool erofs_bh_flush_write_inode(struct erofs_buffer_head *bh)
 {
 	struct erofs_inode *const inode = bh->fsprivate;
+	const u16 icount = EROFS_INODE_XATTR_ICOUNT(inode->xattr_isize);
 	erofs_off_t off = erofs_btell(bh, false);
 	union {
 		struct erofs_inode_compact dic;
@@ -375,6 +377,7 @@ static bool erofs_bh_flush_write_inode(struct erofs_buffer_head *bh)
 	switch (inode->inode_isize) {
 	case sizeof(struct erofs_inode_compact):
 		u.dic.i_format = cpu_to_le16(0 | (inode->datalayout << 1));
+		u.dic.i_xattr_icount = cpu_to_le16(icount);
 		u.dic.i_mode = cpu_to_le16(inode->i_mode);
 		u.dic.i_nlink = cpu_to_le16(inode->i_nlink);
 		u.dic.i_size = cpu_to_le32((u32)inode->i_size);
@@ -404,6 +407,7 @@ static bool erofs_bh_flush_write_inode(struct erofs_buffer_head *bh)
 		break;
 	case sizeof(struct erofs_inode_extended):
 		u.die.i_format = cpu_to_le16(1 | (inode->datalayout << 1));
+		u.die.i_xattr_icount = cpu_to_le16(icount);
 		u.die.i_mode = cpu_to_le16(inode->i_mode);
 		u.die.i_nlink = cpu_to_le32(inode->i_nlink);
 		u.die.i_size = cpu_to_le64(inode->i_size);
@@ -445,6 +449,20 @@ static bool erofs_bh_flush_write_inode(struct erofs_buffer_head *bh)
 		return false;
 	off += inode->inode_isize;
 
+	if (inode->xattr_isize) {
+		char *xattrs = erofs_export_xattr_ibody(&inode->i_xattrs,
+							inode->xattr_isize);
+		if (IS_ERR(xattrs))
+			return false;
+
+		ret = dev_write(xattrs, off, inode->xattr_isize);
+		free(xattrs);
+		if (ret)
+			return false;
+
+		off += inode->xattr_isize;
+	}
+
 	if (inode->extent_isize) {
 		/* write compression metadata */
 		off = Z_EROFS_VLE_EXTENT_ALIGN(off);
@@ -499,8 +517,10 @@ int erofs_prepare_inode_buffer(struct erofs_inode *inode)
 
 	DBG_BUGON(inode->bh || inode->bh_inline);
 
-	inodesize = inode->inode_isize + inode->xattr_isize +
-		    inode->extent_isize;
+	inodesize = inode->inode_isize + inode->xattr_isize;
+	if (inode->extent_isize)
+		inodesize = Z_EROFS_VLE_EXTENT_ALIGN(inodesize) +
+			    inode->extent_isize;
 
 	if (is_inode_layout_compression(inode))
 		goto noinline;
@@ -707,6 +727,8 @@ struct erofs_inode *erofs_new_inode(void)
 	inode->i_count = 1;
 
 	init_list_head(&inode->i_subdirs);
+	init_list_head(&inode->i_xattrs);
+
 	inode->idata_size = 0;
 	inode->xattr_isize = 0;
 	inode->extent_isize = 0;
@@ -795,6 +817,11 @@ struct erofs_inode *erofs_mkfs_build_tree(struct erofs_inode *dir)
 	struct dirent *dp;
 	struct erofs_dentry *d;
 
+	ret = erofs_prepare_xattr_ibody(dir->i_srcpath, &dir->i_xattrs);
+	if (ret < 0)
+		return ERR_PTR(ret);
+	dir->xattr_isize = ret;
+
 	if (!S_ISDIR(dir->i_mode)) {
 		if (S_ISLNK(dir->i_mode)) {
 			char *const symlink = malloc(dir->i_size);
diff --git a/lib/xattr.c b/lib/xattr.c
new file mode 100644
index 0000000..8156f3e
--- /dev/null
+++ b/lib/xattr.c
@@ -0,0 +1,296 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * erofs_utils/lib/xattr.c
+ *
+ * Originally contributed by an anonymous person,
+ * heavily changed by Li Guifu <blucerlee@gmail.com>
+ *                and Gao Xiang <hsiangkao@aol.com>
+ */
+#include <stdlib.h>
+#include <sys/xattr.h>
+#ifdef HAVE_LINUX_XATTR_H
+#include <linux/xattr.h>
+#endif
+#include "erofs/print.h"
+#include "erofs/hashtable.h"
+#include "erofs/xattr.h"
+
+#define EA_HASHTABLE_BITS 16
+
+struct xattr_item {
+	const char *kvbuf;
+	unsigned int hash[2], len[2], count;
+	u8 prefix;
+	struct hlist_node node;
+};
+
+struct inode_xattr_node {
+	struct list_head list;
+	struct xattr_item *item;
+};
+
+static DECLARE_HASHTABLE(ea_hashtable, EA_HASHTABLE_BITS);
+
+static struct xattr_prefix {
+	const char *prefix;
+	u16 prefix_len;
+} prefix[] = {
+	[EROFS_XATTR_INDEX_USER] = {
+		XATTR_USER_PREFIX,
+		XATTR_USER_PREFIX_LEN
+	}, [EROFS_XATTR_INDEX_POSIX_ACL_ACCESS] = {
+		XATTR_NAME_POSIX_ACL_ACCESS,
+		sizeof(XATTR_NAME_POSIX_ACL_ACCESS) - 1
+	}, [EROFS_XATTR_INDEX_POSIX_ACL_DEFAULT] = {
+		XATTR_NAME_POSIX_ACL_DEFAULT,
+		sizeof(XATTR_NAME_POSIX_ACL_DEFAULT) - 1
+	}, [EROFS_XATTR_INDEX_TRUSTED] = {
+		XATTR_TRUSTED_PREFIX,
+		XATTR_TRUSTED_PREFIX_LEN
+	}, [EROFS_XATTR_INDEX_SECURITY] = {
+		XATTR_SECURITY_PREFIX,
+		XATTR_SECURITY_PREFIX_LEN
+	}
+};
+
+static unsigned int BKDRHash(char *str, unsigned int len)
+{
+	const unsigned int seed = 131313;
+	unsigned int hash = 0;
+
+	while (len) {
+		hash = hash * seed + (*str++);
+		--len;
+	}
+	return hash;
+}
+
+static unsigned int xattr_item_hash(u8 prefix, char *buf,
+				    unsigned int len[2], unsigned int hash[2])
+{
+	hash[0] = BKDRHash(buf, len[0]);	/* key */
+	hash[1] = BKDRHash(buf + len[0], len[1]);	/* value */
+
+	return prefix ^ hash[0] ^ hash[1];
+}
+
+static struct xattr_item *xattr_item_get(u8 prefix, char *kvbuf,
+					 unsigned int len[2])
+{
+	struct xattr_item *item;
+	unsigned int hash[2], hkey;
+
+	hkey = xattr_item_hash(prefix, kvbuf, len, hash);
+
+	hash_for_each_possible(ea_hashtable, item, node, hkey) {
+		if (prefix == item->prefix &&
+		    item->len[0] == len[0] && item->len[1] == len[1] &&
+		    item->hash[0] == hash[0] && item->hash[1] == hash[1] &&
+		    !memcmp(kvbuf, item->kvbuf, len[0] + len[1])) {
+			free(kvbuf);
+			++item->count;
+			return item;
+		}
+	}
+
+	item = malloc(sizeof(*item));
+	if (!item) {
+		free(kvbuf);
+		return ERR_PTR(-ENOMEM);
+	}
+	INIT_HLIST_NODE(&item->node);
+	item->count = 1;
+	item->kvbuf = kvbuf;
+	item->len[0] = len[0];
+	item->len[1] = len[1];
+	item->hash[0] = hash[0];
+	item->hash[1] = hash[1];
+	item->prefix = prefix;
+	hash_add(ea_hashtable, &item->node, hkey);
+	return item;
+}
+
+static bool match_prefix(const char *key, u8 *index, u16 *len)
+{
+	struct xattr_prefix *p;
+
+	for (p = prefix; p < prefix + ARRAY_SIZE(prefix); ++p) {
+		if (p->prefix && !strncmp(p->prefix, key, p->prefix_len)) {
+			*len = p->prefix_len;
+			*index = p - prefix;
+			return true;
+		}
+	}
+	return false;
+}
+
+static struct xattr_item *parse_one_xattr(const char *path, const char *key,
+					  unsigned int keylen)
+{
+	ssize_t ret;
+	u8 prefix;
+	u16 prefixlen;
+	unsigned int len[2];
+	char *kvbuf;
+
+	erofs_dbg("parse xattr [%s] of %s", path, key);
+
+	if (!match_prefix(key, &prefix, &prefixlen))
+		return ERR_PTR(-ENODATA);
+
+	DBG_BUGON(keylen < prefixlen);
+
+	/* determine length of the value */
+	ret = lgetxattr(path, key, NULL, 0);
+	if (ret < 0)
+		return ERR_PTR(-errno);
+	len[1] = ret;
+
+	/* allocate key-value buffer */
+	len[0] = keylen - prefixlen;
+
+	kvbuf = malloc(len[0] + len[1]);
+	if (!kvbuf)
+		return ERR_PTR(-ENOMEM);
+
+	memcpy(kvbuf, key + prefixlen, len[0]);
+	if (len[1]) {
+		/* copy value to buffer */
+		ret = lgetxattr(path, key, kvbuf + len[0], len[1]);
+		if (ret < 0) {
+			free(kvbuf);
+			return ERR_PTR(-errno);
+		}
+		if (len[1] != ret) {
+			erofs_err("size of xattr value got changed just now (%u-> %ld)",
+				  len[1], ret);
+			len[1] = ret;
+		}
+	}
+	return xattr_item_get(prefix, kvbuf, len);
+}
+
+static int inode_xattr_add(struct list_head *hlist, struct xattr_item *item)
+{
+	struct inode_xattr_node *node = malloc(sizeof(*node));
+
+	if (!node)
+		return -ENOMEM;
+	init_list_head(&node->list);
+	node->item = item;
+	list_add(&node->list, hlist);
+	return 0;
+}
+
+static int read_xattrs_from_file(const char *path, struct list_head *ixattrs)
+{
+	int ret = 0;
+	char *keylst, *key;
+	ssize_t kllen = llistxattr(path, NULL, 0);
+
+	if (kllen < 0 && errno != ENODATA) {
+		erofs_err("llistxattr to get the size of names for %s failed",
+			  path);
+		return -errno;
+	}
+	if (kllen <= 1)
+		return 0;
+
+	keylst = malloc(kllen);
+	if (!keylst)
+		return -ENOMEM;
+
+	/* copy the list of attribute keys to the buffer.*/
+	kllen = llistxattr(path, keylst, kllen);
+	if (kllen < 0) {
+		erofs_err("llistxattr to get names for %s failed", path);
+		ret = -errno;
+		goto err;
+	}
+
+	/*
+	 * loop over the list of zero terminated strings with the
+	 * attribute keys. Use the remaining buffer length to determine
+	 * the end of the list.
+	 */
+	key = keylst;
+	while (kllen > 0) {
+		unsigned int keylen = strlen(key);
+		struct xattr_item *item = parse_one_xattr(path, key, keylen);
+
+		if (IS_ERR(item)) {
+			ret = PTR_ERR(item);
+			goto err;
+		}
+
+		if (ixattrs) {
+			ret = inode_xattr_add(ixattrs, item);
+			if (ret < 0)
+				goto err;
+		}
+		kllen -= keylen + 1;
+		key += keylen + 1;
+	}
+err:
+	free(keylst);
+	return ret;
+
+}
+
+int erofs_prepare_xattr_ibody(const char *path, struct list_head *ixattrs)
+{
+	int ret;
+	struct inode_xattr_node *node;
+
+	ret = read_xattrs_from_file(path, ixattrs);
+	if (ret < 0)
+		return ret;
+
+	if (list_empty(ixattrs))
+		return 0;
+
+	/* get xattr ibody size */
+	ret = sizeof(struct erofs_xattr_ibody_header);
+	list_for_each_entry(node, ixattrs, list) {
+		const struct xattr_item *item = node->item;
+
+		ret += sizeof(struct erofs_xattr_entry);
+		ret = EROFS_XATTR_ALIGN(ret + item->len[0] + item->len[1]);
+	}
+	return ret;
+}
+
+char *erofs_export_xattr_ibody(struct list_head *ixattrs, unsigned int size)
+{
+	struct inode_xattr_node *node, *n;
+	struct erofs_xattr_ibody_header *header;
+	unsigned int p;
+	char *buf = calloc(1, size);
+
+	if (!buf)
+		return ERR_PTR(-ENOMEM);
+
+	header = (struct erofs_xattr_ibody_header *)buf;
+	header->h_shared_count = 0;
+
+	p = sizeof(struct erofs_xattr_ibody_header);
+	list_for_each_entry_safe(node, n, ixattrs, list) {
+		struct xattr_item *const item = node->item;
+		const struct erofs_xattr_entry entry = {
+			.e_name_index = item->prefix,
+			.e_name_len = item->len[0],
+			.e_value_size = cpu_to_le16(item->len[1])
+		};
+
+		memcpy(buf + p, &entry, sizeof(entry));
+		p += sizeof(struct erofs_xattr_entry);
+		memcpy(buf + p, item->kvbuf, item->len[0] + item->len[1]);
+		p = EROFS_XATTR_ALIGN(p + item->len[0] + item->len[1]);
+
+		list_del(&node->list);
+		free(node);
+	}
+	DBG_BUGON(p > size);
+	return buf;
+}
+
-- 
2.17.1


^ permalink raw reply related	[flat|nested] 12+ messages in thread

* [PATCH 2/2] erofs-utils: introduce shared xattr support
  2019-10-05 14:20       ` [PATCH v3 1/2] erofs-utils: introduce inline " Gao Xiang via Linux-erofs
@ 2019-10-05 14:20         ` Gao Xiang via Linux-erofs
  2019-10-05 16:43           ` Li Guifu
  2019-10-05 16:44         ` [PATCH v3 1/2] erofs-utils: introduce inline " Li Guifu
  1 sibling, 1 reply; 12+ messages in thread
From: Gao Xiang via Linux-erofs @ 2019-10-05 14:20 UTC (permalink / raw)
  To: Li Guifu, Chao Yu, linux-erofs

From: Li Guifu <blucerlee@gmail.com>

Large xattrs or xattrs shared by a lot of files
can be stored in shared xattrs rather than
inlined right after inode.

Signed-off-by: Li Guifu <blucerlee@gmail.com>
Signed-off-by: Gao Xiang <hsiangkao@aol.com>
---
 include/erofs/config.h |   2 +
 include/erofs/xattr.h  |   1 +
 lib/config.c           |   1 +
 lib/xattr.c            | 193 ++++++++++++++++++++++++++++++++++++++++-
 mkfs/main.c            |  10 ++-
 5 files changed, 205 insertions(+), 2 deletions(-)

diff --git a/include/erofs/config.h b/include/erofs/config.h
index 8b09430..fde936c 100644
--- a/include/erofs/config.h
+++ b/include/erofs/config.h
@@ -28,6 +28,8 @@ struct erofs_configure {
 	char *c_compr_alg_master;
 	int c_compr_level_master;
 	int c_force_inodeversion;
+	/* < 0, xattr disabled and INT_MAX, always use inline xattrs */
+	int c_inline_xattr_tolerance;
 };
 
 extern struct erofs_configure cfg;
diff --git a/include/erofs/xattr.h b/include/erofs/xattr.h
index 29df025..3dff1ea 100644
--- a/include/erofs/xattr.h
+++ b/include/erofs/xattr.h
@@ -44,5 +44,6 @@
 
 int erofs_prepare_xattr_ibody(const char *path, struct list_head *ixattrs);
 char *erofs_export_xattr_ibody(struct list_head *ixattrs, unsigned int size);
+int erofs_build_shared_xattrs_from_path(const char *path);
 
 #endif
diff --git a/lib/config.c b/lib/config.c
index 110c8b6..dc10754 100644
--- a/lib/config.c
+++ b/lib/config.c
@@ -23,6 +23,7 @@ void erofs_init_configure(void)
 	cfg.c_compr_level_master = -1;
 	sbi.feature_incompat = EROFS_FEATURE_INCOMPAT_LZ4_0PADDING;
 	cfg.c_force_inodeversion = 0;
+	cfg.c_inline_xattr_tolerance = 2;
 }
 
 void erofs_show_config(void)
diff --git a/lib/xattr.c b/lib/xattr.c
index 8156f3e..781d210 100644
--- a/lib/xattr.c
+++ b/lib/xattr.c
@@ -6,20 +6,26 @@
  * heavily changed by Li Guifu <blucerlee@gmail.com>
  *                and Gao Xiang <hsiangkao@aol.com>
  */
+#define _GNU_SOURCE
+#include <limits.h>
 #include <stdlib.h>
 #include <sys/xattr.h>
 #ifdef HAVE_LINUX_XATTR_H
 #include <linux/xattr.h>
 #endif
+#include <sys/stat.h>
+#include <dirent.h>
 #include "erofs/print.h"
 #include "erofs/hashtable.h"
 #include "erofs/xattr.h"
+#include "erofs/cache.h"
 
 #define EA_HASHTABLE_BITS 16
 
 struct xattr_item {
 	const char *kvbuf;
 	unsigned int hash[2], len[2], count;
+	int shared_xattr_id;
 	u8 prefix;
 	struct hlist_node node;
 };
@@ -31,6 +37,9 @@ struct inode_xattr_node {
 
 static DECLARE_HASHTABLE(ea_hashtable, EA_HASHTABLE_BITS);
 
+static LIST_HEAD(shared_xattrs_list);
+static unsigned int shared_xattrs_count, shared_xattrs_size;
+
 static struct xattr_prefix {
 	const char *prefix;
 	u16 prefix_len;
@@ -105,6 +114,7 @@ static struct xattr_item *xattr_item_get(u8 prefix, char *kvbuf,
 	item->len[1] = len[1];
 	item->hash[0] = hash[0];
 	item->hash[1] = hash[1];
+	item->shared_xattr_id = -1;
 	item->prefix = prefix;
 	hash_add(ea_hashtable, &item->node, hkey);
 	return item;
@@ -152,7 +162,6 @@ static struct xattr_item *parse_one_xattr(const char *path, const char *key,
 	kvbuf = malloc(len[0] + len[1]);
 	if (!kvbuf)
 		return ERR_PTR(-ENOMEM);
-
 	memcpy(kvbuf, key + prefixlen, len[0]);
 	if (len[1]) {
 		/* copy value to buffer */
@@ -182,6 +191,23 @@ static int inode_xattr_add(struct list_head *hlist, struct xattr_item *item)
 	return 0;
 }
 
+static int shared_xattr_add(struct xattr_item *item)
+{
+	struct inode_xattr_node *node = malloc(sizeof(*node));
+
+	if (!node)
+		return -ENOMEM;
+
+	init_list_head(&node->list);
+	node->item = item;
+	list_add(&node->list, &shared_xattrs_list);
+
+	shared_xattrs_size += sizeof(struct erofs_xattr_entry);
+	shared_xattrs_size = EROFS_XATTR_ALIGN(shared_xattrs_size +
+					       item->len[0] + item->len[1]);
+	return ++shared_xattrs_count;
+}
+
 static int read_xattrs_from_file(const char *path, struct list_head *ixattrs)
 {
 	int ret = 0;
@@ -227,6 +253,11 @@ static int read_xattrs_from_file(const char *path, struct list_head *ixattrs)
 			ret = inode_xattr_add(ixattrs, item);
 			if (ret < 0)
 				goto err;
+		} else if (item->count == cfg.c_inline_xattr_tolerance + 1) {
+			ret = shared_xattr_add(item);
+			if (ret < 0)
+				goto err;
+			ret = 0;
 		}
 		kllen -= keylen + 1;
 		key += keylen + 1;
@@ -242,6 +273,10 @@ int erofs_prepare_xattr_ibody(const char *path, struct list_head *ixattrs)
 	int ret;
 	struct inode_xattr_node *node;
 
+	/* check if xattr is disabled */
+	if (cfg.c_inline_xattr_tolerance < 0)
+		return 0;
+
 	ret = read_xattrs_from_file(path, ixattrs);
 	if (ret < 0)
 		return ret;
@@ -254,16 +289,155 @@ int erofs_prepare_xattr_ibody(const char *path, struct list_head *ixattrs)
 	list_for_each_entry(node, ixattrs, list) {
 		const struct xattr_item *item = node->item;
 
+		if (item->shared_xattr_id >= 0) {
+			ret += sizeof(__le32);
+			continue;
+		}
 		ret += sizeof(struct erofs_xattr_entry);
 		ret = EROFS_XATTR_ALIGN(ret + item->len[0] + item->len[1]);
 	}
 	return ret;
 }
 
+static int erofs_count_all_xattrs_from_path(const char *path)
+{
+	int ret;
+	DIR *_dir;
+	struct stat64 st;
+
+	_dir = opendir(path);
+	if (!_dir) {
+		erofs_err("%s, failed to opendir at %s: %s",
+			  __func__, path, erofs_strerror(errno));
+		return -errno;
+	}
+
+	ret = 0;
+	while (1) {
+		struct dirent *dp;
+		char buf[PATH_MAX];
+
+		/*
+		 * set errno to 0 before calling readdir() in order to
+		 * distinguish end of stream and from an error.
+		 */
+		errno = 0;
+		dp = readdir(_dir);
+		if (!dp)
+			break;
+
+		if (is_dot_dotdot(dp->d_name) ||
+		    !strncmp(dp->d_name, "lost+found", strlen("lost+found")))
+			continue;
+
+		ret = snprintf(buf, PATH_MAX, "%s/%s", path, dp->d_name);
+
+		if (ret < 0 || ret >= PATH_MAX) {
+			/* ignore the too long path */
+			ret = -ENOMEM;
+			goto fail;
+		}
+
+		ret = read_xattrs_from_file(buf, NULL);
+		if (ret)
+			goto fail;
+
+		ret = lstat64(buf, &st);
+		if (ret) {
+			ret = -errno;
+			goto fail;
+		}
+
+		if (!S_ISDIR(st.st_mode))
+			continue;
+
+		ret = erofs_count_all_xattrs_from_path(buf);
+		if (ret)
+			goto fail;
+	}
+
+	if (errno)
+		ret = -errno;
+
+fail:
+	closedir(_dir);
+	return ret;
+}
+
+int erofs_build_shared_xattrs_from_path(const char *path)
+{
+	int ret;
+	struct erofs_buffer_head *bh;
+	struct inode_xattr_node *node, *n;
+	char *buf;
+	unsigned int p;
+	erofs_off_t off;
+
+	/* check if xattr or shared xattr is disabled */
+	if (cfg.c_inline_xattr_tolerance < 0 ||
+	    cfg.c_inline_xattr_tolerance == INT_MAX)
+		return 0;
+
+	if (shared_xattrs_size || shared_xattrs_count) {
+		DBG_BUGON(1);
+		return -EINVAL;
+	}
+
+	ret = erofs_count_all_xattrs_from_path(path);
+	if (ret)
+		return ret;
+
+	if (!shared_xattrs_size)
+		return 0;
+
+	buf = malloc(shared_xattrs_size);
+	if (!buf)
+		return -ENOMEM;
+
+	bh = erofs_balloc(XATTR, shared_xattrs_size, 0, 0);
+	if (IS_ERR(bh)) {
+		free(buf);
+		return PTR_ERR(bh);
+	}
+	bh->op = &erofs_skip_write_bhops;
+
+	erofs_mapbh(bh->block, true);
+	off = erofs_btell(bh, false);
+
+	sbi.xattr_blkaddr = off / EROFS_BLKSIZ;
+	off %= EROFS_BLKSIZ;
+	p = 0;
+
+	list_for_each_entry_safe(node, n, &shared_xattrs_list, list) {
+		struct xattr_item *const item = node->item;
+		const struct erofs_xattr_entry entry = {
+			.e_name_index = item->prefix,
+			.e_name_len = item->len[0],
+			.e_value_size = cpu_to_le16(item->len[1])
+		};
+
+		list_del(&node->list);
+
+		item->shared_xattr_id = (off + p) /
+			sizeof(struct erofs_xattr_entry);
+
+		memcpy(buf + p, &entry, sizeof(entry));
+		p += sizeof(struct erofs_xattr_entry);
+		memcpy(buf + p, item->kvbuf, item->len[0] + item->len[1]);
+		p = EROFS_XATTR_ALIGN(p + item->len[0] + item->len[1]);
+		free(node);
+	}
+
+	bh->fsprivate = buf;
+	bh->op = &erofs_buf_write_bhops;
+	return 0;
+}
+
 char *erofs_export_xattr_ibody(struct list_head *ixattrs, unsigned int size)
 {
 	struct inode_xattr_node *node, *n;
 	struct erofs_xattr_ibody_header *header;
+	LIST_HEAD(ilst);
 	unsigned int p;
 	char *buf = calloc(1, size);
 
@@ -276,6 +450,23 @@ char *erofs_export_xattr_ibody(struct list_head *ixattrs, unsigned int size)
 	p = sizeof(struct erofs_xattr_ibody_header);
 	list_for_each_entry_safe(node, n, ixattrs, list) {
 		struct xattr_item *const item = node->item;
+
+		list_del(&node->list);
+
+		/* move inline xattrs to the onstack list */
+		if (item->shared_xattr_id < 0) {
+			list_add(&node->list, &ilst);
+			continue;
+		}
+
+		*(__le32 *)(buf + p) = cpu_to_le32(item->shared_xattr_id);
+		p += sizeof(__le32);
+		++header->h_shared_count;
+		free(node);
+	}
+
+	list_for_each_entry_safe(node, n, &ilst, list) {
+		struct xattr_item *const item = node->item;
 		const struct erofs_xattr_entry entry = {
 			.e_name_index = item->prefix,
 			.e_name_len = item->len[0],
diff --git a/mkfs/main.c b/mkfs/main.c
index 4b279c0..978c5b4 100644
--- a/mkfs/main.c
+++ b/mkfs/main.c
@@ -19,6 +19,7 @@
 #include "erofs/inode.h"
 #include "erofs/io.h"
 #include "erofs/compress.h"
+#include "erofs/xattr.h"
 
 #define EROFS_SUPER_END (EROFS_SUPER_OFFSET + sizeof(struct erofs_super_block))
 
@@ -169,7 +170,7 @@ int erofs_mkfs_update_super_block(struct erofs_buffer_head *bh,
 		.build_time_nsec = cpu_to_le32(sbi.build_time_nsec),
 		.blocks = 0,
 		.meta_blkaddr  = sbi.meta_blkaddr,
-		.xattr_blkaddr = 0,
+		.xattr_blkaddr = sbi.xattr_blkaddr,
 		.feature_incompat = cpu_to_le32(sbi.feature_incompat),
 	};
 	const unsigned int sb_blksize =
@@ -259,6 +260,13 @@ int main(int argc, char **argv)
 
 	erofs_inode_manager_init();
 
+	err = erofs_build_shared_xattrs_from_path(cfg.c_src_path);
+	if (err) {
+		erofs_err("Failed to build shared xattrs: %s",
+			  erofs_strerror(err));
+		goto exit;
+	}
+
 	root_inode = erofs_mkfs_build_tree_from_path(NULL, cfg.c_src_path);
 	if (IS_ERR(root_inode)) {
 		err = PTR_ERR(root_inode);
-- 
2.17.1


^ permalink raw reply related	[flat|nested] 12+ messages in thread

* Re: [PATCH 2/2] erofs-utils: introduce shared xattr support
  2019-10-05 14:20         ` [PATCH 2/2] erofs-utils: introduce shared " Gao Xiang via Linux-erofs
@ 2019-10-05 16:43           ` Li Guifu
  2019-10-06  5:01             ` Gao Xiang via Linux-erofs
  0 siblings, 1 reply; 12+ messages in thread
From: Li Guifu @ 2019-10-05 16:43 UTC (permalink / raw)
  To: Gao Xiang, Li Guifu, Chao Yu, linux-erofs

> From: Li Guifu <blucerlee@gmail.com>
> 
> Large xattrs or xattrs shared by a lot of files
> can be stored in shared xattrs rather than
> inlined right after inode.
> 
> Signed-off-by: Li Guifu <blucerlee@gmail.com>
> Signed-off-by: Gao Xiang <hsiangkao@aol.com>

Dear Gao Xiang,
   Should It need to add a configure parameter to 
cfg.c_inline_xattr_tolerance which is a custome threshold
of shared xattr ?

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: [PATCH v3 1/2] erofs-utils: introduce inline xattr support
  2019-10-05 14:20       ` [PATCH v3 1/2] erofs-utils: introduce inline " Gao Xiang via Linux-erofs
  2019-10-05 14:20         ` [PATCH 2/2] erofs-utils: introduce shared " Gao Xiang via Linux-erofs
@ 2019-10-05 16:44         ` Li Guifu
  1 sibling, 0 replies; 12+ messages in thread
From: Li Guifu @ 2019-10-05 16:44 UTC (permalink / raw)
  To: Gao Xiang, Li Guifu, Chao Yu, linux-erofs; +Cc: htyuxe+dhbrei4sq0df8

> From: "htyuxe+dhbrei4sq0df8@grr.la" <htyuxe+dhbrei4sq0df8@grr.la>
> 
> Load xattrs from source files and pack them into target image.
> 
> Signed-off-by: htyuxe+dhbrei4sq0df8@grr.la <htyuxe+dhbrei4sq0df8@grr.la>
> Signed-off-by: Li Guifu <blucerlee@gmail.com>
> Signed-off-by: Gao Xiang <hsiangkao@aol.com>
> ---
It looks good
Tested-by: Li Guifu <blucerlee@gmail.com>

Thanks,

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: [PATCH 2/2] erofs-utils: introduce shared xattr support
  2019-10-05 16:43           ` Li Guifu
@ 2019-10-06  5:01             ` Gao Xiang via Linux-erofs
  2019-10-14 11:42               ` [PATCH v4 1/2] erofs-utils: introduce inline " Gao Xiang via Linux-erofs
  0 siblings, 1 reply; 12+ messages in thread
From: Gao Xiang via Linux-erofs @ 2019-10-06  5:01 UTC (permalink / raw)
  To: Li Guifu; +Cc: linux-erofs

Hi Guifu,

On Sun, Oct 06, 2019 at 12:43:21AM +0800, Li Guifu wrote:
> > From: Li Guifu <blucerlee@gmail.com>
> > 
> > Large xattrs or xattrs shared by a lot of files
> > can be stored in shared xattrs rather than
> > inlined right after inode.
> > 
> > Signed-off-by: Li Guifu <blucerlee@gmail.com>
> > Signed-off-by: Gao Xiang <hsiangkao@aol.com>
> 
> Dear Gao Xiang,
>   Should It need to add a configure parameter to
> cfg.c_inline_xattr_tolerance which is a custome threshold
> of shared xattr ?

Make sense. I planned to add a brand new command argument but
I need to refer other mkfs first and see if some common name for
this. (If you have time, please help find as well...)

Thanks,
Gao Xiang


^ permalink raw reply	[flat|nested] 12+ messages in thread

* [PATCH v4 1/2] erofs-utils: introduce inline xattr support
  2019-10-06  5:01             ` Gao Xiang via Linux-erofs
@ 2019-10-14 11:42               ` Gao Xiang via Linux-erofs
  2019-10-14 11:42                 ` [PATCH v4 2/2] erofs-utils: introduce shared " Gao Xiang via Linux-erofs
  0 siblings, 1 reply; 12+ messages in thread
From: Gao Xiang via Linux-erofs @ 2019-10-14 11:42 UTC (permalink / raw)
  To: Li Guifu, linux-erofs; +Cc: Miao Xie, htyuxe+dhbrei4sq0df8

From: "htyuxe+dhbrei4sq0df8@grr.la" <htyuxe+dhbrei4sq0df8@grr.la>

Load xattrs from source files and pack them into target image.

Signed-off-by: htyuxe+dhbrei4sq0df8@grr.la <htyuxe+dhbrei4sq0df8@grr.la>
Signed-off-by: Li Guifu <blucerlee@gmail.com>
Signed-off-by: Gao Xiang <hsiangkao@aol.com>
---
changes since v3:
 - add a option "-x" to indicate minimal refcount of a shared xattr or
   xattr disabled;
 - introduce put_xattritem() for unused xattr cleanup.

 configure.ac              |   1 +
 include/erofs/config.h    |   2 +
 include/erofs/defs.h      |   6 +
 include/erofs/hashtable.h | 462 ++++++++++++++++++++++++++++++++++++++
 include/erofs/internal.h  |   2 +-
 include/erofs/xattr.h     |  48 ++++
 lib/Makefile.am           |   3 +-
 lib/config.c              |   1 +
 lib/inode.c               |  31 ++-
 lib/xattr.c               | 309 +++++++++++++++++++++++++
 mkfs/main.c               |  12 +-
 11 files changed, 872 insertions(+), 5 deletions(-)
 create mode 100644 include/erofs/hashtable.h
 create mode 100644 include/erofs/xattr.h
 create mode 100644 lib/xattr.c

diff --git a/configure.ac b/configure.ac
index 07e034e..8c31cb7 100644
--- a/configure.ac
+++ b/configure.ac
@@ -75,6 +75,7 @@ AC_CHECK_HEADERS(m4_flatten([
 	linux/falloc.h
 	linux/fs.h
 	linux/types.h
+	linux/xattr.h
 	limits.h
 	stddef.h
 	stdint.h
diff --git a/include/erofs/config.h b/include/erofs/config.h
index 9711638..2be05ee 100644
--- a/include/erofs/config.h
+++ b/include/erofs/config.h
@@ -28,6 +28,8 @@ struct erofs_configure {
 	char *c_compr_alg_master;
 	int c_compr_level_master;
 	int c_force_inodeversion;
+	/* < 0, xattr disabled and INT_MAX, always use inline xattrs */
+	int c_inline_xattr_tolerance;
 	u64 c_unix_timestamp;
 };
 
diff --git a/include/erofs/defs.h b/include/erofs/defs.h
index db51350..aa127d0 100644
--- a/include/erofs/defs.h
+++ b/include/erofs/defs.h
@@ -152,6 +152,12 @@ typedef int64_t         s64;
 #define BITS_PER_BYTE       8
 #define BITS_TO_LONGS(nr)   DIV_ROUND_UP(nr, BITS_PER_BYTE * sizeof(long))
 
+#ifdef __SIZEOF_LONG__
+#define BITS_PER_LONG (__CHAR_BIT__ * __SIZEOF_LONG__)
+#else
+#define BITS_PER_LONG __WORDSIZE
+#endif
+
 #define BUG_ON(cond)        assert(!(cond))
 
 #ifdef NDEBUG
diff --git a/include/erofs/hashtable.h b/include/erofs/hashtable.h
new file mode 100644
index 0000000..ab57b56
--- /dev/null
+++ b/include/erofs/hashtable.h
@@ -0,0 +1,462 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * erofs_utils/include/erofs/hashtable.h
+ *
+ * Original code taken from 'linux/include/linux/hash{,table}.h'
+ */
+#ifndef __EROFS_HASHTABLE_H
+#define __EROFS_HASHTABLE_H
+
+/*
+ * Fast hashing routine for ints,  longs and pointers.
+ * (C) 2002 Nadia Yvette Chambers, IBM
+ */
+
+/*
+ * Statically sized hash table implementation
+ * (C) 2012  Sasha Levin <levinsasha928@gmail.com>
+ */
+
+#include "defs.h"
+
+/*
+ * The "GOLDEN_RATIO_PRIME" is used in ifs/btrfs/brtfs_inode.h and
+ * fs/inode.c.  It's not actually prime any more (the previous primes
+ * were actively bad for hashing), but the name remains.
+ */
+#if BITS_PER_LONG == 32
+#define GOLDEN_RATIO_PRIME GOLDEN_RATIO_32
+#define hash_long(val, bits) hash_32(val, bits)
+#elif BITS_PER_LONG == 64
+#define hash_long(val, bits) hash_64(val, bits)
+#define GOLDEN_RATIO_PRIME GOLDEN_RATIO_64
+#else
+#error Wordsize not 32 or 64
+#endif
+
+/*
+ * This hash multiplies the input by a large odd number and takes the
+ * high bits.  Since multiplication propagates changes to the most
+ * significant end only, it is essential that the high bits of the
+ * product be used for the hash value.
+ *
+ * Chuck Lever verified the effectiveness of this technique:
+ * http://www.citi.umich.edu/techreports/reports/citi-tr-00-1.pdf
+ *
+ * Although a random odd number will do, it turns out that the golden
+ * ratio phi = (sqrt(5)-1)/2, or its negative, has particularly nice
+ * properties.  (See Knuth vol 3, section 6.4, exercise 9.)
+ *
+ * These are the negative, (1 - phi) = phi**2 = (3 - sqrt(5))/2,
+ * which is very slightly easier to multiply by and makes no
+ * difference to the hash distribution.
+ */
+#define GOLDEN_RATIO_32 0x61C88647
+#define GOLDEN_RATIO_64 0x61C8864680B583EBull
+
+struct hlist_head {
+	struct hlist_node *first;
+};
+
+struct hlist_node {
+	struct hlist_node *next, **pprev;
+};
+
+/*
+ * Architectures might want to move the poison pointer offset
+ * into some well-recognized area such as 0xdead000000000000,
+ * that is also not mappable by user-space exploits:
+ */
+#ifdef CONFIG_ILLEGAL_POINTER_VALUE
+# define POISON_POINTER_DELTA _AC(CONFIG_ILLEGAL_POINTER_VALUE, UL)
+#else
+# define POISON_POINTER_DELTA 0
+#endif
+
+/*
+ * These are non-NULL pointers that will result in page faults
+ * under normal circumstances, used to verify that nobody uses
+ * non-initialized list entries.
+ */
+#define LIST_POISON1 ((void *) 0x00100100 + POISON_POINTER_DELTA)
+#define LIST_POISON2 ((void *) 0x00200200 + POISON_POINTER_DELTA)
+
+/*
+ * Double linked lists with a single pointer list head.
+ * Mostly useful for hash tables where the two pointer list head is
+ * too wasteful.
+ * You lose the ability to access the tail in O(1).
+ */
+
+#define HLIST_HEAD_INIT { .first = NULL }
+#define HLIST_HEAD(name) struct hlist_head name = { .first = NULL }
+#define INIT_HLIST_HEAD(ptr) ((ptr)->first = NULL)
+static inline void INIT_HLIST_NODE(struct hlist_node *h)
+{
+	h->next = NULL;
+	h->pprev = NULL;
+}
+
+static inline int hlist_unhashed(const struct hlist_node *h)
+{
+	return !h->pprev;
+}
+
+static inline int hlist_empty(const struct hlist_head *h)
+{
+	return !h->first;
+}
+
+static inline void __hlist_del(struct hlist_node *n)
+{
+	struct hlist_node *next = n->next;
+	struct hlist_node **pprev = n->pprev;
+
+	*pprev = next;
+	if (next)
+		next->pprev = pprev;
+}
+
+static inline void hlist_del(struct hlist_node *n)
+{
+	__hlist_del(n);
+	n->next = LIST_POISON1;
+	n->pprev = LIST_POISON2;
+}
+
+static inline void hlist_del_init(struct hlist_node *n)
+{
+	if (!hlist_unhashed(n)) {
+		__hlist_del(n);
+		INIT_HLIST_NODE(n);
+	}
+}
+
+static inline void hlist_add_head(struct hlist_node *n, struct hlist_head *h)
+{
+	struct hlist_node *first = h->first;
+
+	n->next = first;
+	if (first)
+		first->pprev = &n->next;
+	h->first = n;
+	n->pprev = &h->first;
+}
+
+/* next must be != NULL */
+static inline void hlist_add_before(struct hlist_node *n,
+					struct hlist_node *next)
+{
+	n->pprev = next->pprev;
+	n->next = next;
+	next->pprev = &n->next;
+	*(n->pprev) = n;
+}
+
+static inline void hlist_add_behind(struct hlist_node *n,
+				    struct hlist_node *prev)
+{
+	n->next = prev->next;
+	prev->next = n;
+	n->pprev = &prev->next;
+
+	if (n->next)
+		n->next->pprev  = &n->next;
+}
+
+/* after that we'll appear to be on some hlist and hlist_del will work */
+static inline void hlist_add_fake(struct hlist_node *n)
+{
+	n->pprev = &n->next;
+}
+
+/*
+ * Move a list from one list head to another. Fixup the pprev
+ * reference of the first entry if it exists.
+ */
+static inline void hlist_move_list(struct hlist_head *old,
+				   struct hlist_head *new)
+{
+	new->first = old->first;
+	if (new->first)
+		new->first->pprev = &new->first;
+	old->first = NULL;
+}
+
+#define hlist_entry(ptr, type, member) container_of(ptr, type, member)
+
+#define hlist_for_each(pos, head) \
+	for (pos = (head)->first; pos; pos = pos->next)
+
+#define hlist_for_each_safe(pos, n, head) \
+	for (pos = (head)->first; pos && ({ n = pos->next; 1; }); \
+	     pos = n)
+
+#define hlist_entry_safe(ptr, type, member) \
+	({ typeof(ptr) ____ptr = (ptr); \
+	   ____ptr ? hlist_entry(____ptr, type, member) : NULL; \
+	})
+
+/**
+ * hlist_for_each_entry	- iterate over list of given type
+ * @pos:the type * to use as a loop cursor.
+ * @head:the head for your list.
+ * @member:the name of the hlist_node within the struct.
+ */
+#define hlist_for_each_entry(pos, head, member)				\
+	for (pos = hlist_entry_safe((head)->first, typeof(*(pos)), member);\
+	     pos;							\
+	     pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member))
+
+/**
+ * hlist_for_each_entry_continue
+ * iterate over a hlist continuing after current point
+ * @pos:the type * to use as a loop cursor.
+ * @member:the name of the hlist_node within the struct.
+ */
+#define hlist_for_each_entry_continue(pos, member)			\
+	for (pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member);\
+	     pos;							\
+	     pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member))
+
+/**
+ * hlist_for_each_entry_from
+ * iterate over a hlist continuing from current point
+ * @pos:	the type * to use as a loop cursor.
+ * @member:	the name of the hlist_node within the struct.
+ */
+#define hlist_for_each_entry_from(pos, member)				\
+	for (; pos;							\
+	     pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member))
+
+/**
+ * hlist_for_each_entry_safe
+ * iterate over list of given type safe against removal of list entry
+ * @pos:the type * to use as a loop cursor.
+ * @n:another &struct hlist_node to use as temporary storage
+ * @head:the head for your list.
+ * @member:the name of the hlist_node within the struct.
+ */
+#define hlist_for_each_entry_safe(pos, n, head, member)			\
+	for (pos = hlist_entry_safe((head)->first, typeof(*pos), member);\
+		pos && ({ n = pos->member.next; 1; });			\
+		pos = hlist_entry_safe(n, typeof(*pos), member))
+
+static inline u32 __hash_32(u32 val)
+{
+	return val * GOLDEN_RATIO_32;
+}
+
+static inline u32 hash_32(u32 val, unsigned int bits)
+{
+	/* High bits are more random, so use them. */
+	return __hash_32(val) >> (32 - bits);
+}
+
+static __always_inline u32 hash_64(u64 val, unsigned int bits)
+{
+#if BITS_PER_LONG == 64
+	/* 64x64-bit multiply is efficient on all 64-bit processors */
+	return val * GOLDEN_RATIO_64 >> (64 - bits);
+#else
+	/* Hash 64 bits using only 32x32-bit multiply. */
+	return hash_32((u32)val ^ __hash_32(val >> 32), bits);
+#endif
+}
+
+/**
+ * ilog2 - log of base 2 of 32-bit or a 64-bit unsigned value
+ * @n - parameter
+ *
+ * constant-capable log of base 2 calculation
+ * - this can be used to initialise global variables from constant data, hence
+ *   the massive ternary operator construction
+ *
+ * selects the appropriately-sized optimised version depending on sizeof(n)
+ */
+#define ilog2(n)				\
+(								\
+	(n) & (1ULL << 63) ? 63 :	\
+	(n) & (1ULL << 62) ? 62 :	\
+	(n) & (1ULL << 61) ? 61 :	\
+	(n) & (1ULL << 60) ? 60 :	\
+	(n) & (1ULL << 59) ? 59 :	\
+	(n) & (1ULL << 58) ? 58 :	\
+	(n) & (1ULL << 57) ? 57 :	\
+	(n) & (1ULL << 56) ? 56 :	\
+	(n) & (1ULL << 55) ? 55 :	\
+	(n) & (1ULL << 54) ? 54 :	\
+	(n) & (1ULL << 53) ? 53 :	\
+	(n) & (1ULL << 52) ? 52 :	\
+	(n) & (1ULL << 51) ? 51 :	\
+	(n) & (1ULL << 50) ? 50 :	\
+	(n) & (1ULL << 49) ? 49 :	\
+	(n) & (1ULL << 48) ? 48 :	\
+	(n) & (1ULL << 47) ? 47 :	\
+	(n) & (1ULL << 46) ? 46 :	\
+	(n) & (1ULL << 45) ? 45 :	\
+	(n) & (1ULL << 44) ? 44 :	\
+	(n) & (1ULL << 43) ? 43 :	\
+	(n) & (1ULL << 42) ? 42 :	\
+	(n) & (1ULL << 41) ? 41 :	\
+	(n) & (1ULL << 40) ? 40 :	\
+	(n) & (1ULL << 39) ? 39 :	\
+	(n) & (1ULL << 38) ? 38 :	\
+	(n) & (1ULL << 37) ? 37 :	\
+	(n) & (1ULL << 36) ? 36 :	\
+	(n) & (1ULL << 35) ? 35 :	\
+	(n) & (1ULL << 34) ? 34 :	\
+	(n) & (1ULL << 33) ? 33 :	\
+	(n) & (1ULL << 32) ? 32 :	\
+	(n) & (1ULL << 31) ? 31 :	\
+	(n) & (1ULL << 30) ? 30 :	\
+	(n) & (1ULL << 29) ? 29 :	\
+	(n) & (1ULL << 28) ? 28 :	\
+	(n) & (1ULL << 27) ? 27 :	\
+	(n) & (1ULL << 26) ? 26 :	\
+	(n) & (1ULL << 25) ? 25 :	\
+	(n) & (1ULL << 24) ? 24 :	\
+	(n) & (1ULL << 23) ? 23 :	\
+	(n) & (1ULL << 22) ? 22 :	\
+	(n) & (1ULL << 21) ? 21 :	\
+	(n) & (1ULL << 20) ? 20 :	\
+	(n) & (1ULL << 19) ? 19 :	\
+	(n) & (1ULL << 18) ? 18 :	\
+	(n) & (1ULL << 17) ? 17 :	\
+	(n) & (1ULL << 16) ? 16 :	\
+	(n) & (1ULL << 15) ? 15 :	\
+	(n) & (1ULL << 14) ? 14 :	\
+	(n) & (1ULL << 13) ? 13 :	\
+	(n) & (1ULL << 12) ? 12 :	\
+	(n) & (1ULL << 11) ? 11 :	\
+	(n) & (1ULL << 10) ? 10 :	\
+	(n) & (1ULL <<  9) ?  9 :	\
+	(n) & (1ULL <<  8) ?  8 :	\
+	(n) & (1ULL <<  7) ?  7 :	\
+	(n) & (1ULL <<  6) ?  6 :	\
+	(n) & (1ULL <<  5) ?  5 :	\
+	(n) & (1ULL <<  4) ?  4 :	\
+	(n) & (1ULL <<  3) ?  3 :	\
+	(n) & (1ULL <<  2) ?  2 :	\
+	(n) & (1ULL <<  1) ?  1 : 0	\
+)
+
+#define DEFINE_HASHTABLE(name, bits)					\
+	struct hlist_head name[1 << (bits)] =				\
+			{ [0 ... ((1 << (bits)) - 1)] = HLIST_HEAD_INIT }
+
+#define DECLARE_HASHTABLE(name, bits)					\
+	struct hlist_head name[1 << (bits)]
+
+#define HASH_SIZE(name) (ARRAY_SIZE(name))
+#define HASH_BITS(name) ilog2(HASH_SIZE(name))
+
+/* Use hash_32 when possible to allow for fast 32bit hashing in 64bit kernels*/
+#define hash_min(val, bits)						\
+	(sizeof(val) <= 4 ? hash_32(val, bits) : hash_long(val, bits))
+
+static inline void __hash_init(struct hlist_head *ht, unsigned int sz)
+{
+	unsigned int i;
+
+	for (i = 0; i < sz; i++)
+		INIT_HLIST_HEAD(&ht[i]);
+}
+
+/**
+ * hash_init - initialize a hash table
+ * @hashtable: hashtable to be initialized
+ *
+ * Calculates the size of the hashtable from the given parameter, otherwise
+ * same as hash_init_size.
+ *
+ * This has to be a macro since HASH_BITS() will not work on pointers since
+ * it calculates the size during preprocessing.
+ */
+#define hash_init(hashtable) __hash_init(hashtable, HASH_SIZE(hashtable))
+
+/**
+ * hash_add - add an object to a hashtable
+ * @hashtable: hashtable to add to
+ * @node: the &struct hlist_node of the object to be added
+ * @key: the key of the object to be added
+ */
+#define hash_add(hashtable, node, key)					\
+	hlist_add_head(node, &hashtable[hash_min(key, HASH_BITS(hashtable))])
+
+/**
+ * hash_hashed - check whether an object is in any hashtable
+ * @node: the &struct hlist_node of the object to be checked
+ */
+static inline bool hash_hashed(struct hlist_node *node)
+{
+	return !hlist_unhashed(node);
+}
+
+static inline bool __hash_empty(struct hlist_head *ht, unsigned int sz)
+{
+	unsigned int i;
+
+	for (i = 0; i < sz; i++)
+		if (!hlist_empty(&ht[i]))
+			return false;
+
+	return true;
+}
+
+/**
+ * hash_empty - check whether a hashtable is empty
+ * @hashtable: hashtable to check
+ *
+ * This has to be a macro since HASH_BITS() will not work on pointers since
+ * it calculates the size during preprocessing.
+ */
+#define hash_empty(hashtable) __hash_empty(hashtable, HASH_SIZE(hashtable))
+
+/**
+ * hash_del - remove an object from a hashtable
+ * @node: &struct hlist_node of the object to remove
+ */
+static inline void hash_del(struct hlist_node *node)
+{
+	hlist_del_init(node);
+}
+
+/**
+ * hash_for_each - iterate over a hashtable
+ * @name: hashtable to iterate
+ * @bkt: integer to use as bucket loop cursor
+ * @obj: the type * to use as a loop cursor for each entry
+ * @member: the name of the hlist_node within the struct
+ */
+#define hash_for_each(name, bkt, obj, member)				\
+	for ((bkt) = 0, obj = NULL; obj == NULL && (bkt) < HASH_SIZE(name);\
+			(bkt)++)\
+		hlist_for_each_entry(obj, &name[bkt], member)
+
+/**
+ * hash_for_each_safe - iterate over a hashtable safe against removal of
+ * hash entry
+ * @name: hashtable to iterate
+ * @bkt: integer to use as bucket loop cursor
+ * @tmp: a &struct used for temporary storage
+ * @obj: the type * to use as a loop cursor for each entry
+ * @member: the name of the hlist_node within the struct
+ */
+#define hash_for_each_safe(name, bkt, tmp, obj, member)			\
+	for ((bkt) = 0, obj = NULL; obj == NULL && (bkt) < HASH_SIZE(name);\
+			(bkt)++)\
+		hlist_for_each_entry_safe(obj, tmp, &name[bkt], member)
+
+/**
+ * hash_for_each_possible - iterate over all possible objects hashing to the
+ * same bucket
+ * @name: hashtable to iterate
+ * @obj: the type * to use as a loop cursor for each entry
+ * @member: the name of the hlist_node within the struct
+ * @key: the key of the objects to iterate over
+ */
+#define hash_for_each_possible(name, obj, member, key)			\
+	hlist_for_each_entry(obj, &name[hash_min(key, HASH_BITS(name))], member)
+
+#endif
diff --git a/include/erofs/internal.h b/include/erofs/internal.h
index 5384946..25ce7b5 100644
--- a/include/erofs/internal.h
+++ b/include/erofs/internal.h
@@ -61,7 +61,7 @@ struct erofs_sb_info {
 extern struct erofs_sb_info sbi;
 
 struct erofs_inode {
-	struct list_head i_hash, i_subdirs;
+	struct list_head i_hash, i_subdirs, i_xattrs;
 
 	unsigned int i_count;
 	struct erofs_inode *i_parent;
diff --git a/include/erofs/xattr.h b/include/erofs/xattr.h
new file mode 100644
index 0000000..29df025
--- /dev/null
+++ b/include/erofs/xattr.h
@@ -0,0 +1,48 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * erofs_utils/include/erofs/xattr.h
+ *
+ * Originally contributed by an anonymous person,
+ * heavily changed by Li Guifu <blucerlee@gmail.com>
+ *                and Gao Xiang <xiang@kernel.org>
+ */
+#ifndef __EROFS_XATTR_H
+#define __EROFS_XATTR_H
+
+#include "internal.h"
+
+#define EROFS_INODE_XATTR_ICOUNT(_size)	({\
+	u32 __size = le16_to_cpu(_size); \
+	((__size) == 0) ? 0 : \
+	(_size - sizeof(struct erofs_xattr_ibody_header)) / \
+	sizeof(struct erofs_xattr_entry) + 1; })
+
+#ifndef XATTR_USER_PREFIX
+#define XATTR_USER_PREFIX	"user."
+#endif
+#ifndef XATTR_USER_PREFIX_LEN
+#define XATTR_USER_PREFIX_LEN (sizeof(XATTR_USER_PREFIX) - 1)
+#endif
+#ifndef XATTR_SECURITY_PREFIX
+#define XATTR_SECURITY_PREFIX	"security."
+#endif
+#ifndef XATTR_SECURITY_PREFIX_LEN
+#define XATTR_SECURITY_PREFIX_LEN (sizeof(XATTR_SECURITY_PREFIX) - 1)
+#endif
+#ifndef XATTR_TRUSTED_PREFIX
+#define XATTR_TRUSTED_PREFIX	"trusted."
+#endif
+#ifndef XATTR_TRUSTED_PREFIX_LEN
+#define XATTR_TRUSTED_PREFIX_LEN (sizeof(XATTR_TRUSTED_PREFIX) - 1)
+#endif
+#ifndef XATTR_NAME_POSIX_ACL_ACCESS
+#define XATTR_NAME_POSIX_ACL_ACCESS "system.posix_acl_access"
+#endif
+#ifndef XATTR_NAME_POSIX_ACL_DEFAULT
+#define XATTR_NAME_POSIX_ACL_DEFAULT "system.posix_acl_default"
+#endif
+
+int erofs_prepare_xattr_ibody(const char *path, struct list_head *ixattrs);
+char *erofs_export_xattr_ibody(struct list_head *ixattrs, unsigned int size);
+
+#endif
diff --git a/lib/Makefile.am b/lib/Makefile.am
index dea82f7..1ff81f9 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -2,7 +2,8 @@
 # Makefile.am
 
 noinst_LTLIBRARIES = liberofs.la
-liberofs_la_SOURCES = config.c io.c cache.c inode.c compress.c compressor.c
+liberofs_la_SOURCES = config.c io.c cache.c inode.c xattr.c \
+		      compress.c compressor.c
 liberofs_la_CFLAGS = -Wall -Werror -I$(top_srcdir)/include
 if ENABLE_LZ4
 liberofs_la_CFLAGS += ${LZ4_CFLAGS}
diff --git a/lib/config.c b/lib/config.c
index 46625d7..cb42706 100644
--- a/lib/config.c
+++ b/lib/config.c
@@ -21,6 +21,7 @@ void erofs_init_configure(void)
 	cfg.c_dry_run  = false;
 	cfg.c_compr_level_master = -1;
 	cfg.c_force_inodeversion = 0;
+	cfg.c_inline_xattr_tolerance = 1;
 	cfg.c_unix_timestamp = -1;
 }
 
diff --git a/lib/inode.c b/lib/inode.c
index 395caae..b7121e0 100644
--- a/lib/inode.c
+++ b/lib/inode.c
@@ -20,6 +20,7 @@
 #include "erofs/cache.h"
 #include "erofs/io.h"
 #include "erofs/compress.h"
+#include "erofs/xattr.h"
 
 struct erofs_sb_info sbi;
 
@@ -365,6 +366,7 @@ fail:
 static bool erofs_bh_flush_write_inode(struct erofs_buffer_head *bh)
 {
 	struct erofs_inode *const inode = bh->fsprivate;
+	const u16 icount = EROFS_INODE_XATTR_ICOUNT(inode->xattr_isize);
 	erofs_off_t off = erofs_btell(bh, false);
 	union {
 		struct erofs_inode_compact dic;
@@ -375,6 +377,7 @@ static bool erofs_bh_flush_write_inode(struct erofs_buffer_head *bh)
 	switch (inode->inode_isize) {
 	case sizeof(struct erofs_inode_compact):
 		u.dic.i_format = cpu_to_le16(0 | (inode->datalayout << 1));
+		u.dic.i_xattr_icount = cpu_to_le16(icount);
 		u.dic.i_mode = cpu_to_le16(inode->i_mode);
 		u.dic.i_nlink = cpu_to_le16(inode->i_nlink);
 		u.dic.i_size = cpu_to_le32((u32)inode->i_size);
@@ -404,6 +407,7 @@ static bool erofs_bh_flush_write_inode(struct erofs_buffer_head *bh)
 		break;
 	case sizeof(struct erofs_inode_extended):
 		u.die.i_format = cpu_to_le16(1 | (inode->datalayout << 1));
+		u.die.i_xattr_icount = cpu_to_le16(icount);
 		u.die.i_mode = cpu_to_le16(inode->i_mode);
 		u.die.i_nlink = cpu_to_le32(inode->i_nlink);
 		u.die.i_size = cpu_to_le64(inode->i_size);
@@ -445,6 +449,20 @@ static bool erofs_bh_flush_write_inode(struct erofs_buffer_head *bh)
 		return false;
 	off += inode->inode_isize;
 
+	if (inode->xattr_isize) {
+		char *xattrs = erofs_export_xattr_ibody(&inode->i_xattrs,
+							inode->xattr_isize);
+		if (IS_ERR(xattrs))
+			return false;
+
+		ret = dev_write(xattrs, off, inode->xattr_isize);
+		free(xattrs);
+		if (ret)
+			return false;
+
+		off += inode->xattr_isize;
+	}
+
 	if (inode->extent_isize) {
 		/* write compression metadata */
 		off = Z_EROFS_VLE_EXTENT_ALIGN(off);
@@ -499,8 +517,10 @@ int erofs_prepare_inode_buffer(struct erofs_inode *inode)
 
 	DBG_BUGON(inode->bh || inode->bh_inline);
 
-	inodesize = inode->inode_isize + inode->xattr_isize +
-		    inode->extent_isize;
+	inodesize = inode->inode_isize + inode->xattr_isize;
+	if (inode->extent_isize)
+		inodesize = Z_EROFS_VLE_EXTENT_ALIGN(inodesize) +
+			    inode->extent_isize;
 
 	if (is_inode_layout_compression(inode))
 		goto noinline;
@@ -707,6 +727,8 @@ struct erofs_inode *erofs_new_inode(void)
 	inode->i_count = 1;
 
 	init_list_head(&inode->i_subdirs);
+	init_list_head(&inode->i_xattrs);
+
 	inode->idata_size = 0;
 	inode->xattr_isize = 0;
 	inode->extent_isize = 0;
@@ -795,6 +817,11 @@ struct erofs_inode *erofs_mkfs_build_tree(struct erofs_inode *dir)
 	struct dirent *dp;
 	struct erofs_dentry *d;
 
+	ret = erofs_prepare_xattr_ibody(dir->i_srcpath, &dir->i_xattrs);
+	if (ret < 0)
+		return ERR_PTR(ret);
+	dir->xattr_isize = ret;
+
 	if (!S_ISDIR(dir->i_mode)) {
 		if (S_ISLNK(dir->i_mode)) {
 			char *const symlink = malloc(dir->i_size);
diff --git a/lib/xattr.c b/lib/xattr.c
new file mode 100644
index 0000000..d07d325
--- /dev/null
+++ b/lib/xattr.c
@@ -0,0 +1,309 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * erofs_utils/lib/xattr.c
+ *
+ * Originally contributed by an anonymous person,
+ * heavily changed by Li Guifu <blucerlee@gmail.com>
+ *                and Gao Xiang <hsiangkao@aol.com>
+ */
+#include <stdlib.h>
+#include <sys/xattr.h>
+#ifdef HAVE_LINUX_XATTR_H
+#include <linux/xattr.h>
+#endif
+#include "erofs/print.h"
+#include "erofs/hashtable.h"
+#include "erofs/xattr.h"
+
+#define EA_HASHTABLE_BITS 16
+
+struct xattr_item {
+	const char *kvbuf;
+	unsigned int hash[2], len[2], count;
+	u8 prefix;
+	struct hlist_node node;
+};
+
+struct inode_xattr_node {
+	struct list_head list;
+	struct xattr_item *item;
+};
+
+static DECLARE_HASHTABLE(ea_hashtable, EA_HASHTABLE_BITS);
+
+static struct xattr_prefix {
+	const char *prefix;
+	u16 prefix_len;
+} xattr_types[] = {
+	[EROFS_XATTR_INDEX_USER] = {
+		XATTR_USER_PREFIX,
+		XATTR_USER_PREFIX_LEN
+	}, [EROFS_XATTR_INDEX_POSIX_ACL_ACCESS] = {
+		XATTR_NAME_POSIX_ACL_ACCESS,
+		sizeof(XATTR_NAME_POSIX_ACL_ACCESS) - 1
+	}, [EROFS_XATTR_INDEX_POSIX_ACL_DEFAULT] = {
+		XATTR_NAME_POSIX_ACL_DEFAULT,
+		sizeof(XATTR_NAME_POSIX_ACL_DEFAULT) - 1
+	}, [EROFS_XATTR_INDEX_TRUSTED] = {
+		XATTR_TRUSTED_PREFIX,
+		XATTR_TRUSTED_PREFIX_LEN
+	}, [EROFS_XATTR_INDEX_SECURITY] = {
+		XATTR_SECURITY_PREFIX,
+		XATTR_SECURITY_PREFIX_LEN
+	}
+};
+
+static unsigned int BKDRHash(char *str, unsigned int len)
+{
+	const unsigned int seed = 131313;
+	unsigned int hash = 0;
+
+	while (len) {
+		hash = hash * seed + (*str++);
+		--len;
+	}
+	return hash;
+}
+
+static unsigned int xattr_item_hash(u8 prefix, char *buf,
+				    unsigned int len[2], unsigned int hash[2])
+{
+	hash[0] = BKDRHash(buf, len[0]);	/* key */
+	hash[1] = BKDRHash(buf + len[0], len[1]);	/* value */
+
+	return prefix ^ hash[0] ^ hash[1];
+}
+
+static unsigned int put_xattritem(struct xattr_item *item)
+{
+	if (item->count > 1)
+		return --item->count;
+	free(item);
+	return 0;
+}
+
+static struct xattr_item *get_xattritem(u8 prefix, char *kvbuf,
+					unsigned int len[2])
+{
+	struct xattr_item *item;
+	unsigned int hash[2], hkey;
+
+	hkey = xattr_item_hash(prefix, kvbuf, len, hash);
+
+	hash_for_each_possible(ea_hashtable, item, node, hkey) {
+		if (prefix == item->prefix &&
+		    item->len[0] == len[0] && item->len[1] == len[1] &&
+		    item->hash[0] == hash[0] && item->hash[1] == hash[1] &&
+		    !memcmp(kvbuf, item->kvbuf, len[0] + len[1])) {
+			free(kvbuf);
+			++item->count;
+			return item;
+		}
+	}
+
+	item = malloc(sizeof(*item));
+	if (!item) {
+		free(kvbuf);
+		return ERR_PTR(-ENOMEM);
+	}
+	INIT_HLIST_NODE(&item->node);
+	item->count = 1;
+	item->kvbuf = kvbuf;
+	item->len[0] = len[0];
+	item->len[1] = len[1];
+	item->hash[0] = hash[0];
+	item->hash[1] = hash[1];
+	item->prefix = prefix;
+	hash_add(ea_hashtable, &item->node, hkey);
+	return item;
+}
+
+static bool match_prefix(const char *key, u8 *index, u16 *len)
+{
+	struct xattr_prefix *p;
+
+	for (p = xattr_types; p < xattr_types + ARRAY_SIZE(xattr_types); ++p) {
+		if (p->prefix && !strncmp(p->prefix, key, p->prefix_len)) {
+			*len = p->prefix_len;
+			*index = p - xattr_types;
+			return true;
+		}
+	}
+	return false;
+}
+
+static struct xattr_item *parse_one_xattr(const char *path, const char *key,
+					  unsigned int keylen)
+{
+	ssize_t ret;
+	u8 prefix;
+	u16 prefixlen;
+	unsigned int len[2];
+	char *kvbuf;
+
+	erofs_dbg("parse xattr [%s] of %s", path, key);
+
+	if (!match_prefix(key, &prefix, &prefixlen))
+		return ERR_PTR(-ENODATA);
+
+	DBG_BUGON(keylen < prefixlen);
+
+	/* determine length of the value */
+	ret = lgetxattr(path, key, NULL, 0);
+	if (ret < 0)
+		return ERR_PTR(-errno);
+	len[1] = ret;
+
+	/* allocate key-value buffer */
+	len[0] = keylen - prefixlen;
+
+	kvbuf = malloc(len[0] + len[1]);
+	if (!kvbuf)
+		return ERR_PTR(-ENOMEM);
+
+	memcpy(kvbuf, key + prefixlen, len[0]);
+	if (len[1]) {
+		/* copy value to buffer */
+		ret = lgetxattr(path, key, kvbuf + len[0], len[1]);
+		if (ret < 0) {
+			free(kvbuf);
+			return ERR_PTR(-errno);
+		}
+		if (len[1] != ret) {
+			erofs_err("size of xattr value got changed just now (%u-> %ld)",
+				  len[1], (long)ret);
+			len[1] = ret;
+		}
+	}
+	return get_xattritem(prefix, kvbuf, len);
+}
+
+static int inode_xattr_add(struct list_head *hlist, struct xattr_item *item)
+{
+	struct inode_xattr_node *node = malloc(sizeof(*node));
+
+	if (!node)
+		return -ENOMEM;
+	init_list_head(&node->list);
+	node->item = item;
+	list_add(&node->list, hlist);
+	return 0;
+}
+
+static int read_xattrs_from_file(const char *path, struct list_head *ixattrs)
+{
+	int ret = 0;
+	char *keylst, *key;
+	ssize_t kllen = llistxattr(path, NULL, 0);
+
+	if (kllen < 0 && errno != ENODATA) {
+		erofs_err("llistxattr to get the size of names for %s failed",
+			  path);
+		return -errno;
+	}
+	if (kllen <= 1)
+		return 0;
+
+	keylst = malloc(kllen);
+	if (!keylst)
+		return -ENOMEM;
+
+	/* copy the list of attribute keys to the buffer.*/
+	kllen = llistxattr(path, keylst, kllen);
+	if (kllen < 0) {
+		erofs_err("llistxattr to get names for %s failed", path);
+		ret = -errno;
+		goto err;
+	}
+
+	/*
+	 * loop over the list of zero terminated strings with the
+	 * attribute keys. Use the remaining buffer length to determine
+	 * the end of the list.
+	 */
+	key = keylst;
+	while (kllen > 0) {
+		unsigned int keylen = strlen(key);
+		struct xattr_item *item = parse_one_xattr(path, key, keylen);
+
+		if (IS_ERR(item)) {
+			ret = PTR_ERR(item);
+			goto err;
+		}
+
+		if (ixattrs) {
+			ret = inode_xattr_add(ixattrs, item);
+			if (ret < 0)
+				goto err;
+		}
+		kllen -= keylen + 1;
+		key += keylen + 1;
+	}
+err:
+	free(keylst);
+	return ret;
+
+}
+
+int erofs_prepare_xattr_ibody(const char *path, struct list_head *ixattrs)
+{
+	int ret;
+	struct inode_xattr_node *node;
+
+	/* check if xattr is disabled */
+	if (cfg.c_inline_xattr_tolerance < 0)
+		return 0;
+
+	ret = read_xattrs_from_file(path, ixattrs);
+	if (ret < 0)
+		return ret;
+
+	if (list_empty(ixattrs))
+		return 0;
+
+	/* get xattr ibody size */
+	ret = sizeof(struct erofs_xattr_ibody_header);
+	list_for_each_entry(node, ixattrs, list) {
+		const struct xattr_item *item = node->item;
+
+		ret += sizeof(struct erofs_xattr_entry);
+		ret = EROFS_XATTR_ALIGN(ret + item->len[0] + item->len[1]);
+	}
+	return ret;
+}
+
+char *erofs_export_xattr_ibody(struct list_head *ixattrs, unsigned int size)
+{
+	struct inode_xattr_node *node, *n;
+	struct erofs_xattr_ibody_header *header;
+	unsigned int p;
+	char *buf = calloc(1, size);
+
+	if (!buf)
+		return ERR_PTR(-ENOMEM);
+
+	header = (struct erofs_xattr_ibody_header *)buf;
+	header->h_shared_count = 0;
+
+	p = sizeof(struct erofs_xattr_ibody_header);
+	list_for_each_entry_safe(node, n, ixattrs, list) {
+		struct xattr_item *const item = node->item;
+		const struct erofs_xattr_entry entry = {
+			.e_name_index = item->prefix,
+			.e_name_len = item->len[0],
+			.e_value_size = cpu_to_le16(item->len[1])
+		};
+
+		memcpy(buf + p, &entry, sizeof(entry));
+		p += sizeof(struct erofs_xattr_entry);
+		memcpy(buf + p, item->kvbuf, item->len[0] + item->len[1]);
+		p = EROFS_XATTR_ALIGN(p + item->len[0] + item->len[1]);
+
+		list_del(&node->list);
+		free(node);
+		put_xattritem(item);
+	}
+	DBG_BUGON(p > size);
+	return buf;
+}
+
diff --git a/mkfs/main.c b/mkfs/main.c
index 536b784..0df2a96 100644
--- a/mkfs/main.c
+++ b/mkfs/main.c
@@ -28,6 +28,7 @@ static void usage(void)
 	fprintf(stderr, "Generate erofs image from DIRECTORY to FILE, and [options] are:\n");
 	fprintf(stderr, " -zX[,Y]   X=compressor (Y=compression level, optional)\n");
 	fprintf(stderr, " -d#       set output message level to # (maximum 9)\n");
+	fprintf(stderr, " -x#       set xattr tolerance to # (< 0, disable xattrs; default 1)\n");
 	fprintf(stderr, " -EX[,...] X=extended options\n");
 	fprintf(stderr, " -T#       set a fixed UNIX timestamp # to all files\n");
 }
@@ -94,7 +95,7 @@ static int mkfs_parse_options_cfg(int argc, char *argv[])
 	char *endptr;
 	int opt, i;
 
-	while ((opt = getopt(argc, argv, "d:z:E:T:")) != -1) {
+	while ((opt = getopt(argc, argv, "d:x:z:E:T:")) != -1) {
 		switch (opt) {
 		case 'z':
 			if (!optarg) {
@@ -122,6 +123,15 @@ static int mkfs_parse_options_cfg(int argc, char *argv[])
 			cfg.c_dbg_lvl = i;
 			break;
 
+		case 'x':
+			i = strtol(optarg, &endptr, 0);
+			if (*endptr != '\0') {
+				erofs_err("invalid xattr tolerance %s", optarg);
+				return -EINVAL;
+			}
+			cfg.c_inline_xattr_tolerance = i;
+			break;
+
 		case 'E':
 			opt = parse_extended_opts(optarg);
 			if (opt)
-- 
2.17.1


^ permalink raw reply related	[flat|nested] 12+ messages in thread

* [PATCH v4 2/2] erofs-utils: introduce shared xattr support
  2019-10-14 11:42               ` [PATCH v4 1/2] erofs-utils: introduce inline " Gao Xiang via Linux-erofs
@ 2019-10-14 11:42                 ` Gao Xiang via Linux-erofs
  2019-10-14 23:53                   ` [PATCH v5 " Gao Xiang via Linux-erofs
  0 siblings, 1 reply; 12+ messages in thread
From: Gao Xiang via Linux-erofs @ 2019-10-14 11:42 UTC (permalink / raw)
  To: Li Guifu, linux-erofs; +Cc: Miao Xie

From: Li Guifu <blucerlee@gmail.com>

Large xattrs or xattrs shared by a lot of files
can be stored in shared xattrs rather than
inlined right after inode.

Signed-off-by: Li Guifu <blucerlee@gmail.com>
Signed-off-by: Gao Xiang <hsiangkao@aol.com>
---
 include/erofs/defs.h  |   2 +-
 include/erofs/xattr.h |   1 +
 lib/config.c          |   2 +-
 lib/inode.c           |   1 -
 lib/xattr.c           | 208 +++++++++++++++++++++++++++++++++++++++++-
 mkfs/main.c           |  12 ++-
 6 files changed, 220 insertions(+), 6 deletions(-)

diff --git a/include/erofs/defs.h b/include/erofs/defs.h
index aa127d0..c67035d 100644
--- a/include/erofs/defs.h
+++ b/include/erofs/defs.h
@@ -14,7 +14,7 @@
 #include <stdint.h>
 #include <assert.h>
 #include <inttypes.h>
-
+#include <limits.h>
 #include <stdbool.h>
 
 #ifdef HAVE_CONFIG_H
diff --git a/include/erofs/xattr.h b/include/erofs/xattr.h
index 29df025..3dff1ea 100644
--- a/include/erofs/xattr.h
+++ b/include/erofs/xattr.h
@@ -44,5 +44,6 @@
 
 int erofs_prepare_xattr_ibody(const char *path, struct list_head *ixattrs);
 char *erofs_export_xattr_ibody(struct list_head *ixattrs, unsigned int size);
+int erofs_build_shared_xattrs_from_path(const char *path);
 
 #endif
diff --git a/lib/config.c b/lib/config.c
index cb42706..cbbecce 100644
--- a/lib/config.c
+++ b/lib/config.c
@@ -21,7 +21,7 @@ void erofs_init_configure(void)
 	cfg.c_dry_run  = false;
 	cfg.c_compr_level_master = -1;
 	cfg.c_force_inodeversion = 0;
-	cfg.c_inline_xattr_tolerance = 1;
+	cfg.c_inline_xattr_tolerance = 2;
 	cfg.c_unix_timestamp = -1;
 }
 
diff --git a/lib/inode.c b/lib/inode.c
index b7121e0..86c465e 100644
--- a/lib/inode.c
+++ b/lib/inode.c
@@ -8,7 +8,6 @@
  * with heavy changes by Gao Xiang <gaoxiang25@huawei.com>
  */
 #define _GNU_SOURCE
-#include <limits.h>
 #include <string.h>
 #include <stdlib.h>
 #include <stdio.h>
diff --git a/lib/xattr.c b/lib/xattr.c
index d07d325..18a8c82 100644
--- a/lib/xattr.c
+++ b/lib/xattr.c
@@ -6,20 +6,25 @@
  * heavily changed by Li Guifu <blucerlee@gmail.com>
  *                and Gao Xiang <hsiangkao@aol.com>
  */
+#define _GNU_SOURCE
 #include <stdlib.h>
 #include <sys/xattr.h>
 #ifdef HAVE_LINUX_XATTR_H
 #include <linux/xattr.h>
 #endif
+#include <sys/stat.h>
+#include <dirent.h>
 #include "erofs/print.h"
 #include "erofs/hashtable.h"
 #include "erofs/xattr.h"
+#include "erofs/cache.h"
 
 #define EA_HASHTABLE_BITS 16
 
 struct xattr_item {
 	const char *kvbuf;
 	unsigned int hash[2], len[2], count;
+	int shared_xattr_id;
 	u8 prefix;
 	struct hlist_node node;
 };
@@ -31,6 +36,9 @@ struct inode_xattr_node {
 
 static DECLARE_HASHTABLE(ea_hashtable, EA_HASHTABLE_BITS);
 
+static LIST_HEAD(shared_xattrs_list);
+static unsigned int shared_xattrs_count, shared_xattrs_size;
+
 static struct xattr_prefix {
 	const char *prefix;
 	u16 prefix_len;
@@ -113,6 +121,7 @@ static struct xattr_item *get_xattritem(u8 prefix, char *kvbuf,
 	item->len[1] = len[1];
 	item->hash[0] = hash[0];
 	item->hash[1] = hash[1];
+	item->shared_xattr_id = -1;
 	item->prefix = prefix;
 	hash_add(ea_hashtable, &item->node, hkey);
 	return item;
@@ -160,7 +169,6 @@ static struct xattr_item *parse_one_xattr(const char *path, const char *key,
 	kvbuf = malloc(len[0] + len[1]);
 	if (!kvbuf)
 		return ERR_PTR(-ENOMEM);
-
 	memcpy(kvbuf, key + prefixlen, len[0]);
 	if (len[1]) {
 		/* copy value to buffer */
@@ -190,6 +198,23 @@ static int inode_xattr_add(struct list_head *hlist, struct xattr_item *item)
 	return 0;
 }
 
+static int shared_xattr_add(struct xattr_item *item)
+{
+	struct inode_xattr_node *node = malloc(sizeof(*node));
+
+	if (!node)
+		return -ENOMEM;
+
+	init_list_head(&node->list);
+	node->item = item;
+	list_add(&node->list, &shared_xattrs_list);
+
+	shared_xattrs_size += sizeof(struct erofs_xattr_entry);
+	shared_xattrs_size = EROFS_XATTR_ALIGN(shared_xattrs_size +
+					       item->len[0] + item->len[1]);
+	return ++shared_xattrs_count;
+}
+
 static int read_xattrs_from_file(const char *path, struct list_head *ixattrs)
 {
 	int ret = 0;
@@ -235,6 +260,11 @@ static int read_xattrs_from_file(const char *path, struct list_head *ixattrs)
 			ret = inode_xattr_add(ixattrs, item);
 			if (ret < 0)
 				goto err;
+		} else if (item->count == cfg.c_inline_xattr_tolerance + 1) {
+			ret = shared_xattr_add(item);
+			if (ret < 0)
+				goto err;
+			ret = 0;
 		}
 		kllen -= keylen + 1;
 		key += keylen + 1;
@@ -266,16 +296,174 @@ int erofs_prepare_xattr_ibody(const char *path, struct list_head *ixattrs)
 	list_for_each_entry(node, ixattrs, list) {
 		const struct xattr_item *item = node->item;
 
+		if (item->shared_xattr_id >= 0) {
+			ret += sizeof(__le32);
+			continue;
+		}
 		ret += sizeof(struct erofs_xattr_entry);
 		ret = EROFS_XATTR_ALIGN(ret + item->len[0] + item->len[1]);
 	}
 	return ret;
 }
 
+static int erofs_count_all_xattrs_from_path(const char *path)
+{
+	int ret;
+	DIR *_dir;
+	struct stat64 st;
+
+	_dir = opendir(path);
+	if (!_dir) {
+		erofs_err("%s, failed to opendir at %s: %s",
+			  __func__, path, erofs_strerror(errno));
+		return -errno;
+	}
+
+	ret = 0;
+	while (1) {
+		struct dirent *dp;
+		char buf[PATH_MAX];
+
+		/*
+		 * set errno to 0 before calling readdir() in order to
+		 * distinguish end of stream and from an error.
+		 */
+		errno = 0;
+		dp = readdir(_dir);
+		if (!dp)
+			break;
+
+		if (is_dot_dotdot(dp->d_name) ||
+		    !strncmp(dp->d_name, "lost+found", strlen("lost+found")))
+			continue;
+
+		ret = snprintf(buf, PATH_MAX, "%s/%s", path, dp->d_name);
+
+		if (ret < 0 || ret >= PATH_MAX) {
+			/* ignore the too long path */
+			ret = -ENOMEM;
+			goto fail;
+		}
+
+		ret = read_xattrs_from_file(buf, NULL);
+		if (ret)
+			goto fail;
+
+		ret = lstat64(buf, &st);
+		if (ret) {
+			ret = -errno;
+			goto fail;
+		}
+
+		if (!S_ISDIR(st.st_mode))
+			continue;
+
+		ret = erofs_count_all_xattrs_from_path(buf);
+		if (ret)
+			goto fail;
+	}
+
+	if (errno)
+		ret = -errno;
+
+fail:
+	closedir(_dir);
+	return ret;
+}
+
+static void erofs_cleanxattrs(bool sharedxattrs)
+{
+	unsigned int i;
+	struct xattr_item *item;
+
+	hash_for_each(ea_hashtable, i, item, node) {
+		if (sharedxattrs && item->shared_xattr_id >= 0)
+			continue;
+
+		hash_del(&item->node);
+		free(item);
+	}
+
+	if (sharedxattrs)
+		return;
+
+	shared_xattrs_size = shared_xattrs_count = 0;
+}
+
+int erofs_build_shared_xattrs_from_path(const char *path)
+{
+	int ret;
+	struct erofs_buffer_head *bh;
+	struct inode_xattr_node *node, *n;
+	char *buf;
+	unsigned int p;
+	erofs_off_t off;
+
+	/* check if xattr or shared xattr is disabled */
+	if (cfg.c_inline_xattr_tolerance < 0 ||
+	    cfg.c_inline_xattr_tolerance == INT_MAX)
+		return 0;
+
+	if (shared_xattrs_size || shared_xattrs_count) {
+		DBG_BUGON(1);
+		return -EINVAL;
+	}
+
+	ret = erofs_count_all_xattrs_from_path(path);
+	if (ret)
+		return ret;
+
+	if (!shared_xattrs_size)
+		return 0;
+
+	buf = malloc(shared_xattrs_size);
+	if (!buf)
+		return -ENOMEM;
+
+	bh = erofs_balloc(XATTR, shared_xattrs_size, 0, 0);
+	if (IS_ERR(bh)) {
+		free(buf);
+		return PTR_ERR(bh);
+	}
+	bh->op = &erofs_skip_write_bhops;
+
+	erofs_mapbh(bh->block, true);
+	off = erofs_btell(bh, false);
+
+	sbi.xattr_blkaddr = off / EROFS_BLKSIZ;
+	off %= EROFS_BLKSIZ;
+	p = 0;
+
+	list_for_each_entry_safe(node, n, &shared_xattrs_list, list) {
+		struct xattr_item *const item = node->item;
+		const struct erofs_xattr_entry entry = {
+			.e_name_index = item->prefix,
+			.e_name_len = item->len[0],
+			.e_value_size = cpu_to_le16(item->len[1])
+		};
+
+		list_del(&node->list);
+
+		item->shared_xattr_id = (off + p) /
+			sizeof(struct erofs_xattr_entry);
+
+		memcpy(buf + p, &entry, sizeof(entry));
+		p += sizeof(struct erofs_xattr_entry);
+		memcpy(buf + p, item->kvbuf, item->len[0] + item->len[1]);
+		p = EROFS_XATTR_ALIGN(p + item->len[0] + item->len[1]);
+		free(node);
+	}
+	bh->fsprivate = buf;
+	bh->op = &erofs_buf_write_bhops;
+	erofs_cleanxattrs(true);
+	return 0;
+}
+
 char *erofs_export_xattr_ibody(struct list_head *ixattrs, unsigned int size)
 {
 	struct inode_xattr_node *node, *n;
 	struct erofs_xattr_ibody_header *header;
+	LIST_HEAD(ilst);
 	unsigned int p;
 	char *buf = calloc(1, size);
 
@@ -288,6 +476,24 @@ char *erofs_export_xattr_ibody(struct list_head *ixattrs, unsigned int size)
 	p = sizeof(struct erofs_xattr_ibody_header);
 	list_for_each_entry_safe(node, n, ixattrs, list) {
 		struct xattr_item *const item = node->item;
+
+		list_del(&node->list);
+
+		/* move inline xattrs to the onstack list */
+		if (item->shared_xattr_id < 0) {
+			list_add(&node->list, &ilst);
+			continue;
+		}
+
+		*(__le32 *)(buf + p) = cpu_to_le32(item->shared_xattr_id);
+		p += sizeof(__le32);
+		++header->h_shared_count;
+		free(node);
+		put_xattritem(item);
+	}
+
+	list_for_each_entry_safe(node, n, &ilst, list) {
+		struct xattr_item *const item = node->item;
 		const struct erofs_xattr_entry entry = {
 			.e_name_index = item->prefix,
 			.e_name_len = item->len[0],
diff --git a/mkfs/main.c b/mkfs/main.c
index 0df2a96..71c81f5 100644
--- a/mkfs/main.c
+++ b/mkfs/main.c
@@ -19,6 +19,7 @@
 #include "erofs/inode.h"
 #include "erofs/io.h"
 #include "erofs/compress.h"
+#include "erofs/xattr.h"
 
 #define EROFS_SUPER_END (EROFS_SUPER_OFFSET + sizeof(struct erofs_super_block))
 
@@ -28,7 +29,7 @@ static void usage(void)
 	fprintf(stderr, "Generate erofs image from DIRECTORY to FILE, and [options] are:\n");
 	fprintf(stderr, " -zX[,Y]   X=compressor (Y=compression level, optional)\n");
 	fprintf(stderr, " -d#       set output message level to # (maximum 9)\n");
-	fprintf(stderr, " -x#       set xattr tolerance to # (< 0, disable xattrs; default 1)\n");
+	fprintf(stderr, " -x#       set xattr tolerance to # (< 0, disable xattrs; default 2)\n");
 	fprintf(stderr, " -EX[,...] X=extended options\n");
 	fprintf(stderr, " -T#       set a fixed UNIX timestamp # to all files\n");
 }
@@ -188,7 +189,7 @@ int erofs_mkfs_update_super_block(struct erofs_buffer_head *bh,
 		.build_time_nsec = cpu_to_le32(sbi.build_time_nsec),
 		.blocks = 0,
 		.meta_blkaddr  = sbi.meta_blkaddr,
-		.xattr_blkaddr = 0,
+		.xattr_blkaddr = sbi.xattr_blkaddr,
 		.feature_incompat = cpu_to_le32(sbi.feature_incompat),
 	};
 	const unsigned int sb_blksize =
@@ -284,6 +285,13 @@ int main(int argc, char **argv)
 
 	erofs_inode_manager_init();
 
+	err = erofs_build_shared_xattrs_from_path(cfg.c_src_path);
+	if (err) {
+		erofs_err("Failed to build shared xattrs: %s",
+			  erofs_strerror(err));
+		goto exit;
+	}
+
 	root_inode = erofs_mkfs_build_tree_from_path(NULL, cfg.c_src_path);
 	if (IS_ERR(root_inode)) {
 		err = PTR_ERR(root_inode);
-- 
2.17.1


^ permalink raw reply related	[flat|nested] 12+ messages in thread

* [PATCH v5 2/2] erofs-utils: introduce shared xattr support
  2019-10-14 11:42                 ` [PATCH v4 2/2] erofs-utils: introduce shared " Gao Xiang via Linux-erofs
@ 2019-10-14 23:53                   ` Gao Xiang via Linux-erofs
  0 siblings, 0 replies; 12+ messages in thread
From: Gao Xiang via Linux-erofs @ 2019-10-14 23:53 UTC (permalink / raw)
  To: Li Guifu, linux-erofs; +Cc: Miao Xie

From: Li Guifu <blucerlee@gmail.com>

Large xattrs or xattrs shared by a lot of files
can be stored in shared xattrs rather than
inlined right after inode.

Signed-off-by: Li Guifu <blucerlee@gmail.com>
Signed-off-by: Gao Xiang <hsiangkao@aol.com>
---
Changes since v4:
 - cleanxattrs when !shared_xattrs_size as well

 include/erofs/defs.h  |   2 +-
 include/erofs/xattr.h |   1 +
 lib/config.c          |   2 +-
 lib/inode.c           |   1 -
 lib/xattr.c           | 209 +++++++++++++++++++++++++++++++++++++++++-
 mkfs/main.c           |  12 ++-
 6 files changed, 221 insertions(+), 6 deletions(-)

diff --git a/include/erofs/defs.h b/include/erofs/defs.h
index aa127d0..c67035d 100644
--- a/include/erofs/defs.h
+++ b/include/erofs/defs.h
@@ -14,7 +14,7 @@
 #include <stdint.h>
 #include <assert.h>
 #include <inttypes.h>
-
+#include <limits.h>
 #include <stdbool.h>
 
 #ifdef HAVE_CONFIG_H
diff --git a/include/erofs/xattr.h b/include/erofs/xattr.h
index 29df025..3dff1ea 100644
--- a/include/erofs/xattr.h
+++ b/include/erofs/xattr.h
@@ -44,5 +44,6 @@
 
 int erofs_prepare_xattr_ibody(const char *path, struct list_head *ixattrs);
 char *erofs_export_xattr_ibody(struct list_head *ixattrs, unsigned int size);
+int erofs_build_shared_xattrs_from_path(const char *path);
 
 #endif
diff --git a/lib/config.c b/lib/config.c
index cb42706..cbbecce 100644
--- a/lib/config.c
+++ b/lib/config.c
@@ -21,7 +21,7 @@ void erofs_init_configure(void)
 	cfg.c_dry_run  = false;
 	cfg.c_compr_level_master = -1;
 	cfg.c_force_inodeversion = 0;
-	cfg.c_inline_xattr_tolerance = 1;
+	cfg.c_inline_xattr_tolerance = 2;
 	cfg.c_unix_timestamp = -1;
 }
 
diff --git a/lib/inode.c b/lib/inode.c
index b7121e0..86c465e 100644
--- a/lib/inode.c
+++ b/lib/inode.c
@@ -8,7 +8,6 @@
  * with heavy changes by Gao Xiang <gaoxiang25@huawei.com>
  */
 #define _GNU_SOURCE
-#include <limits.h>
 #include <string.h>
 #include <stdlib.h>
 #include <stdio.h>
diff --git a/lib/xattr.c b/lib/xattr.c
index d07d325..1564016 100644
--- a/lib/xattr.c
+++ b/lib/xattr.c
@@ -6,20 +6,25 @@
  * heavily changed by Li Guifu <blucerlee@gmail.com>
  *                and Gao Xiang <hsiangkao@aol.com>
  */
+#define _GNU_SOURCE
 #include <stdlib.h>
 #include <sys/xattr.h>
 #ifdef HAVE_LINUX_XATTR_H
 #include <linux/xattr.h>
 #endif
+#include <sys/stat.h>
+#include <dirent.h>
 #include "erofs/print.h"
 #include "erofs/hashtable.h"
 #include "erofs/xattr.h"
+#include "erofs/cache.h"
 
 #define EA_HASHTABLE_BITS 16
 
 struct xattr_item {
 	const char *kvbuf;
 	unsigned int hash[2], len[2], count;
+	int shared_xattr_id;
 	u8 prefix;
 	struct hlist_node node;
 };
@@ -31,6 +36,9 @@ struct inode_xattr_node {
 
 static DECLARE_HASHTABLE(ea_hashtable, EA_HASHTABLE_BITS);
 
+static LIST_HEAD(shared_xattrs_list);
+static unsigned int shared_xattrs_count, shared_xattrs_size;
+
 static struct xattr_prefix {
 	const char *prefix;
 	u16 prefix_len;
@@ -113,6 +121,7 @@ static struct xattr_item *get_xattritem(u8 prefix, char *kvbuf,
 	item->len[1] = len[1];
 	item->hash[0] = hash[0];
 	item->hash[1] = hash[1];
+	item->shared_xattr_id = -1;
 	item->prefix = prefix;
 	hash_add(ea_hashtable, &item->node, hkey);
 	return item;
@@ -160,7 +169,6 @@ static struct xattr_item *parse_one_xattr(const char *path, const char *key,
 	kvbuf = malloc(len[0] + len[1]);
 	if (!kvbuf)
 		return ERR_PTR(-ENOMEM);
-
 	memcpy(kvbuf, key + prefixlen, len[0]);
 	if (len[1]) {
 		/* copy value to buffer */
@@ -190,6 +198,23 @@ static int inode_xattr_add(struct list_head *hlist, struct xattr_item *item)
 	return 0;
 }
 
+static int shared_xattr_add(struct xattr_item *item)
+{
+	struct inode_xattr_node *node = malloc(sizeof(*node));
+
+	if (!node)
+		return -ENOMEM;
+
+	init_list_head(&node->list);
+	node->item = item;
+	list_add(&node->list, &shared_xattrs_list);
+
+	shared_xattrs_size += sizeof(struct erofs_xattr_entry);
+	shared_xattrs_size = EROFS_XATTR_ALIGN(shared_xattrs_size +
+					       item->len[0] + item->len[1]);
+	return ++shared_xattrs_count;
+}
+
 static int read_xattrs_from_file(const char *path, struct list_head *ixattrs)
 {
 	int ret = 0;
@@ -235,6 +260,11 @@ static int read_xattrs_from_file(const char *path, struct list_head *ixattrs)
 			ret = inode_xattr_add(ixattrs, item);
 			if (ret < 0)
 				goto err;
+		} else if (item->count == cfg.c_inline_xattr_tolerance + 1) {
+			ret = shared_xattr_add(item);
+			if (ret < 0)
+				goto err;
+			ret = 0;
 		}
 		kllen -= keylen + 1;
 		key += keylen + 1;
@@ -266,16 +296,175 @@ int erofs_prepare_xattr_ibody(const char *path, struct list_head *ixattrs)
 	list_for_each_entry(node, ixattrs, list) {
 		const struct xattr_item *item = node->item;
 
+		if (item->shared_xattr_id >= 0) {
+			ret += sizeof(__le32);
+			continue;
+		}
 		ret += sizeof(struct erofs_xattr_entry);
 		ret = EROFS_XATTR_ALIGN(ret + item->len[0] + item->len[1]);
 	}
 	return ret;
 }
 
+static int erofs_count_all_xattrs_from_path(const char *path)
+{
+	int ret;
+	DIR *_dir;
+	struct stat64 st;
+
+	_dir = opendir(path);
+	if (!_dir) {
+		erofs_err("%s, failed to opendir at %s: %s",
+			  __func__, path, erofs_strerror(errno));
+		return -errno;
+	}
+
+	ret = 0;
+	while (1) {
+		struct dirent *dp;
+		char buf[PATH_MAX];
+
+		/*
+		 * set errno to 0 before calling readdir() in order to
+		 * distinguish end of stream and from an error.
+		 */
+		errno = 0;
+		dp = readdir(_dir);
+		if (!dp)
+			break;
+
+		if (is_dot_dotdot(dp->d_name) ||
+		    !strncmp(dp->d_name, "lost+found", strlen("lost+found")))
+			continue;
+
+		ret = snprintf(buf, PATH_MAX, "%s/%s", path, dp->d_name);
+
+		if (ret < 0 || ret >= PATH_MAX) {
+			/* ignore the too long path */
+			ret = -ENOMEM;
+			goto fail;
+		}
+
+		ret = read_xattrs_from_file(buf, NULL);
+		if (ret)
+			goto fail;
+
+		ret = lstat64(buf, &st);
+		if (ret) {
+			ret = -errno;
+			goto fail;
+		}
+
+		if (!S_ISDIR(st.st_mode))
+			continue;
+
+		ret = erofs_count_all_xattrs_from_path(buf);
+		if (ret)
+			goto fail;
+	}
+
+	if (errno)
+		ret = -errno;
+
+fail:
+	closedir(_dir);
+	return ret;
+}
+
+static void erofs_cleanxattrs(bool sharedxattrs)
+{
+	unsigned int i;
+	struct xattr_item *item;
+
+	hash_for_each(ea_hashtable, i, item, node) {
+		if (sharedxattrs && item->shared_xattr_id >= 0)
+			continue;
+
+		hash_del(&item->node);
+		free(item);
+	}
+
+	if (sharedxattrs)
+		return;
+
+	shared_xattrs_size = shared_xattrs_count = 0;
+}
+
+int erofs_build_shared_xattrs_from_path(const char *path)
+{
+	int ret;
+	struct erofs_buffer_head *bh;
+	struct inode_xattr_node *node, *n;
+	char *buf;
+	unsigned int p;
+	erofs_off_t off;
+
+	/* check if xattr or shared xattr is disabled */
+	if (cfg.c_inline_xattr_tolerance < 0 ||
+	    cfg.c_inline_xattr_tolerance == INT_MAX)
+		return 0;
+
+	if (shared_xattrs_size || shared_xattrs_count) {
+		DBG_BUGON(1);
+		return -EINVAL;
+	}
+
+	ret = erofs_count_all_xattrs_from_path(path);
+	if (ret)
+		return ret;
+
+	if (!shared_xattrs_size)
+		goto out;
+
+	buf = malloc(shared_xattrs_size);
+	if (!buf)
+		return -ENOMEM;
+
+	bh = erofs_balloc(XATTR, shared_xattrs_size, 0, 0);
+	if (IS_ERR(bh)) {
+		free(buf);
+		return PTR_ERR(bh);
+	}
+	bh->op = &erofs_skip_write_bhops;
+
+	erofs_mapbh(bh->block, true);
+	off = erofs_btell(bh, false);
+
+	sbi.xattr_blkaddr = off / EROFS_BLKSIZ;
+	off %= EROFS_BLKSIZ;
+	p = 0;
+
+	list_for_each_entry_safe(node, n, &shared_xattrs_list, list) {
+		struct xattr_item *const item = node->item;
+		const struct erofs_xattr_entry entry = {
+			.e_name_index = item->prefix,
+			.e_name_len = item->len[0],
+			.e_value_size = cpu_to_le16(item->len[1])
+		};
+
+		list_del(&node->list);
+
+		item->shared_xattr_id = (off + p) /
+			sizeof(struct erofs_xattr_entry);
+
+		memcpy(buf + p, &entry, sizeof(entry));
+		p += sizeof(struct erofs_xattr_entry);
+		memcpy(buf + p, item->kvbuf, item->len[0] + item->len[1]);
+		p = EROFS_XATTR_ALIGN(p + item->len[0] + item->len[1]);
+		free(node);
+	}
+	bh->fsprivate = buf;
+	bh->op = &erofs_buf_write_bhops;
+out:
+	erofs_cleanxattrs(true);
+	return 0;
+}
+
 char *erofs_export_xattr_ibody(struct list_head *ixattrs, unsigned int size)
 {
 	struct inode_xattr_node *node, *n;
 	struct erofs_xattr_ibody_header *header;
+	LIST_HEAD(ilst);
 	unsigned int p;
 	char *buf = calloc(1, size);
 
@@ -288,6 +477,24 @@ char *erofs_export_xattr_ibody(struct list_head *ixattrs, unsigned int size)
 	p = sizeof(struct erofs_xattr_ibody_header);
 	list_for_each_entry_safe(node, n, ixattrs, list) {
 		struct xattr_item *const item = node->item;
+
+		list_del(&node->list);
+
+		/* move inline xattrs to the onstack list */
+		if (item->shared_xattr_id < 0) {
+			list_add(&node->list, &ilst);
+			continue;
+		}
+
+		*(__le32 *)(buf + p) = cpu_to_le32(item->shared_xattr_id);
+		p += sizeof(__le32);
+		++header->h_shared_count;
+		free(node);
+		put_xattritem(item);
+	}
+
+	list_for_each_entry_safe(node, n, &ilst, list) {
+		struct xattr_item *const item = node->item;
 		const struct erofs_xattr_entry entry = {
 			.e_name_index = item->prefix,
 			.e_name_len = item->len[0],
diff --git a/mkfs/main.c b/mkfs/main.c
index 0df2a96..71c81f5 100644
--- a/mkfs/main.c
+++ b/mkfs/main.c
@@ -19,6 +19,7 @@
 #include "erofs/inode.h"
 #include "erofs/io.h"
 #include "erofs/compress.h"
+#include "erofs/xattr.h"
 
 #define EROFS_SUPER_END (EROFS_SUPER_OFFSET + sizeof(struct erofs_super_block))
 
@@ -28,7 +29,7 @@ static void usage(void)
 	fprintf(stderr, "Generate erofs image from DIRECTORY to FILE, and [options] are:\n");
 	fprintf(stderr, " -zX[,Y]   X=compressor (Y=compression level, optional)\n");
 	fprintf(stderr, " -d#       set output message level to # (maximum 9)\n");
-	fprintf(stderr, " -x#       set xattr tolerance to # (< 0, disable xattrs; default 1)\n");
+	fprintf(stderr, " -x#       set xattr tolerance to # (< 0, disable xattrs; default 2)\n");
 	fprintf(stderr, " -EX[,...] X=extended options\n");
 	fprintf(stderr, " -T#       set a fixed UNIX timestamp # to all files\n");
 }
@@ -188,7 +189,7 @@ int erofs_mkfs_update_super_block(struct erofs_buffer_head *bh,
 		.build_time_nsec = cpu_to_le32(sbi.build_time_nsec),
 		.blocks = 0,
 		.meta_blkaddr  = sbi.meta_blkaddr,
-		.xattr_blkaddr = 0,
+		.xattr_blkaddr = sbi.xattr_blkaddr,
 		.feature_incompat = cpu_to_le32(sbi.feature_incompat),
 	};
 	const unsigned int sb_blksize =
@@ -284,6 +285,13 @@ int main(int argc, char **argv)
 
 	erofs_inode_manager_init();
 
+	err = erofs_build_shared_xattrs_from_path(cfg.c_src_path);
+	if (err) {
+		erofs_err("Failed to build shared xattrs: %s",
+			  erofs_strerror(err));
+		goto exit;
+	}
+
 	root_inode = erofs_mkfs_build_tree_from_path(NULL, cfg.c_src_path);
 	if (IS_ERR(root_inode)) {
 		err = PTR_ERR(root_inode);
-- 
2.17.1


^ permalink raw reply related	[flat|nested] 12+ messages in thread

end of thread, other threads:[~2019-10-14 23:53 UTC | newest]

Thread overview: 12+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-07-26 11:50 [PATCH] erofs-utils: introduce xattr support htyuxe+dhbrei4sq0df8
2019-08-05 14:54 ` Li Guifu
2019-08-05 17:30   ` Gao Xiang
2019-08-11 17:10     ` [PATCH v2] erofs-utils: introduce preliminary " Gao Xiang
2019-10-05 14:20       ` [PATCH v3 1/2] erofs-utils: introduce inline " Gao Xiang via Linux-erofs
2019-10-05 14:20         ` [PATCH 2/2] erofs-utils: introduce shared " Gao Xiang via Linux-erofs
2019-10-05 16:43           ` Li Guifu
2019-10-06  5:01             ` Gao Xiang via Linux-erofs
2019-10-14 11:42               ` [PATCH v4 1/2] erofs-utils: introduce inline " Gao Xiang via Linux-erofs
2019-10-14 11:42                 ` [PATCH v4 2/2] erofs-utils: introduce shared " Gao Xiang via Linux-erofs
2019-10-14 23:53                   ` [PATCH v5 " Gao Xiang via Linux-erofs
2019-10-05 16:44         ` [PATCH v3 1/2] erofs-utils: introduce inline " Li Guifu

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).