[RFC PATCH 0/3] erofs-utils: compressed fragments feature

[RFC PATCH 0/3] erofs-utils: compressed fragments feature

[Yue Hu]

In order to achieve greater compression ratio, let's introduce
compressed fragments feature which can merge tail of per-file or the
whole files into one special inode to reach the target.

And we can also set pcluster size to fragments inode for different
compression requirments.

In this patchset, we also improve the uncompressed data layout of
compressed files. Just write it from 'clusterofs' instead of 0 since it
can benefit from in-place I/O. For now, it only goes with fragments.

The main idea above is from Xiang.

Here is some test data of Linux 5.10.87 source code under Ubuntu 18.04:

linux-5.10.87 (erofs, uncompressed)                1.1G

linux-5.10.87 (erofs, lz4hc,12 4k fragments,4k)    304M
linux-5.10.87 (erofs, lz4hc,12 8k fragments,8k)    271M
linux-5.10.87 (erofs, lz4hc,12 16k fragments,16k)  245M
linux-5.10.87 (erofs, lz4hc,12 32k fragments,32k)  228M
linux-5.10.87 (erofs, lz4hc,12 64k fragments,64k)  220M

linux-5.10.87 (erofs, lz4hc,12 4k vanilla)         396M
linux-5.10.87 (erofs, lz4hc,12 8k vanilla)         376M
linux-5.10.87 (erofs, lz4hc,12 16k vanilla)        364M
linux-5.10.87 (erofs, lz4hc,12 32k vanilla)        359M
linux-5.10.87 (erofs, lz4hc,12 64k vanilla)        358M

Usage:
mkfs.erofs -zlz4hc,12 -C65536 -Efragments,65536 foo.erofs.img foo/

Yue Hu (3):
  erofs-utils: lib: add support for fragments data decompression
  erofs-utils: lib: support on-disk offset for shifted decompression
  erofs-utils: introduce compressed fragments support

 include/erofs/compress.h   |  2 +-
 include/erofs/config.h     |  3 +-
 include/erofs/decompress.h |  3 ++
 include/erofs/fragments.h  | 25 ++++++++++
 include/erofs/inode.h      |  2 +
 include/erofs/internal.h   |  9 ++++
 include/erofs_fs.h         | 20 ++++++--
 lib/Makefile.am            |  4 +-
 lib/compress.c             | 94 ++++++++++++++++++++++++++++----------
 lib/data.c                 | 33 ++++++++++++-
 lib/decompress.c           | 10 +++-
 lib/fragments.c            | 77 +++++++++++++++++++++++++++++++
 lib/inode.c                | 50 +++++++++++++-------
 lib/super.c                |  1 +
 lib/zmap.c                 | 14 ++++++
 mkfs/main.c                | 63 ++++++++++++++++++++++---
 16 files changed, 352 insertions(+), 58 deletions(-)
 create mode 100644 include/erofs/fragments.h
 create mode 100644 lib/fragments.c

-- 
2.17.1

[RFC PATCH 1/3] erofs-utils: lib: add support for fragments data decompression

[Yue Hu]

Add compressed fragments support for erofsfuse.

Signed-off-by: Yue Hu <huyue2@coolpad.com>
---
 include/erofs/internal.h |  8 ++++++++
 include/erofs_fs.h       | 15 +++++++++++----
 lib/data.c               | 26 ++++++++++++++++++++++++++
 lib/zmap.c               | 14 ++++++++++++++
 4 files changed, 59 insertions(+), 4 deletions(-)

diff --git a/include/erofs/internal.h b/include/erofs/internal.h
index 6a70f11..129ea54 100644
--- a/include/erofs/internal.h
+++ b/include/erofs/internal.h
@@ -102,6 +102,7 @@ struct erofs_sb_info {
 		u16 devt_slotoff;		/* used for mkfs */
 		u16 device_id_mask;		/* used for others */
 	};
+	erofs_nid_t fragments_nid;
 };
 
 /* global sbi */
@@ -132,6 +133,7 @@ EROFS_FEATURE_FUNCS(big_pcluster, incompat, INCOMPAT_BIG_PCLUSTER)
 EROFS_FEATURE_FUNCS(chunked_file, incompat, INCOMPAT_CHUNKED_FILE)
 EROFS_FEATURE_FUNCS(device_table, incompat, INCOMPAT_DEVICE_TABLE)
 EROFS_FEATURE_FUNCS(ztailpacking, incompat, INCOMPAT_ZTAILPACKING)
+EROFS_FEATURE_FUNCS(fragments, incompat, INCOMPAT_FRAGMENTS)
 EROFS_FEATURE_FUNCS(sb_chksum, compat, COMPAT_SB_CHKSUM)
 
 #define EROFS_I_EA_INITED	(1 << 0)
@@ -190,6 +192,8 @@ struct erofs_inode {
 	void *eof_tailraw;
 	unsigned int eof_tailrawsize;
 
+	erofs_off_t fragmentoff;
+
 	union {
 		void *compressmeta;
 		void *chunkindexes;
@@ -201,6 +205,7 @@ struct erofs_inode {
 			uint64_t z_tailextent_headlcn;
 			unsigned int    z_idataoff;
 #define z_idata_size	idata_size
+#define z_fragmentoff	fragmentoff
 		};
 	};
 #ifdef WITH_ANDROID
@@ -276,6 +281,7 @@ enum {
 	BH_Mapped,
 	BH_Encoded,
 	BH_FullMapped,
+	BH_Fragments,
 };
 
 /* Has a disk mapping */
@@ -286,6 +292,8 @@ enum {
 #define EROFS_MAP_ENCODED	(1 << BH_Encoded)
 /* The length of extent is full */
 #define EROFS_MAP_FULL_MAPPED	(1 << BH_FullMapped)
+/* Located in fragments */
+#define EROFS_MAP_FRAGMENTS	(1 << BH_Fragments)
 
 struct erofs_map_blocks {
 	char mpage[EROFS_BLKSIZ];
diff --git a/include/erofs_fs.h b/include/erofs_fs.h
index 08f9761..4a0c74b 100644
--- a/include/erofs_fs.h
+++ b/include/erofs_fs.h
@@ -25,13 +25,15 @@
 #define EROFS_FEATURE_INCOMPAT_CHUNKED_FILE	0x00000004
 #define EROFS_FEATURE_INCOMPAT_DEVICE_TABLE	0x00000008
 #define EROFS_FEATURE_INCOMPAT_ZTAILPACKING	0x00000010
+#define EROFS_FEATURE_INCOMPAT_FRAGMENTS	0x00000020
 #define EROFS_ALL_FEATURE_INCOMPAT		\
 	(EROFS_FEATURE_INCOMPAT_LZ4_0PADDING | \
 	 EROFS_FEATURE_INCOMPAT_COMPR_CFGS | \
 	 EROFS_FEATURE_INCOMPAT_BIG_PCLUSTER | \
 	 EROFS_FEATURE_INCOMPAT_CHUNKED_FILE | \
 	 EROFS_FEATURE_INCOMPAT_DEVICE_TABLE | \
-	 EROFS_FEATURE_INCOMPAT_ZTAILPACKING)
+	 EROFS_FEATURE_INCOMPAT_ZTAILPACKING | \
+	 EROFS_FEATURE_INCOMPAT_FRAGMENTS)
 
 #define EROFS_SB_EXTSLOT_SIZE	16
 
@@ -294,16 +296,21 @@ struct z_erofs_lzma_cfgs {
  * bit 1 : HEAD1 big pcluster (0 - off; 1 - on)
  * bit 2 : HEAD2 big pcluster (0 - off; 1 - on)
  * bit 3 : tailpacking inline pcluster (0 - off; 1 - on)
+ * bit 4 : fragment pcluster (0 - off; 1 - on)
  */
 #define Z_EROFS_ADVISE_COMPACTED_2B		0x0001
 #define Z_EROFS_ADVISE_BIG_PCLUSTER_1		0x0002
 #define Z_EROFS_ADVISE_BIG_PCLUSTER_2		0x0004
 #define Z_EROFS_ADVISE_INLINE_PCLUSTER		0x0008
+#define Z_EROFS_ADVISE_FRAGMENT_PCLUSTER	0x0010
 
 struct z_erofs_map_header {
-	__le16	h_reserved1;
-	/* record the size of tailpacking data */
-	__le16  h_idata_size;
+	union {
+		/* direct addressing for fragment offset */
+		__le32	h_fragmentoff;
+		/* record the size of tailpacking data */
+		__le16	h_idata_size;
+	};
 	__le16	h_advise;
 	/*
 	 * bit 0-3 : algorithm type of head 1 (logical cluster type 01);
diff --git a/lib/data.c b/lib/data.c
index 6bc554d..c2ecbc9 100644
--- a/lib/data.c
+++ b/lib/data.c
@@ -275,6 +275,32 @@ static int z_erofs_read_data(struct erofs_inode *inode, char *buffer,
 			continue;
 		}
 
+		if (map.m_flags & EROFS_MAP_FRAGMENTS) {
+			struct erofs_inode fragments = {
+					.nid = sbi.fragments_nid,
+			};
+			char *out;
+
+			ret = erofs_read_inode_from_disk(&fragments);
+			if (ret)
+				break;
+			out = malloc(length);
+			if (!out) {
+				ret = -ENOMEM;
+				break;
+			}
+			ret = z_erofs_read_data(&fragments, out, length,
+						inode->z_fragmentoff);
+			if (ret < 0) {
+				free(out);
+				break;
+			}
+			memcpy(buffer + end - offset, out + skip, length -
+			       skip);
+			free(out);
+			continue;
+		}
+
 		if (map.m_plen > bufsize) {
 			bufsize = map.m_plen;
 			raw = realloc(raw, bufsize);
diff --git a/lib/zmap.c b/lib/zmap.c
index 95745c5..2b45a4f 100644
--- a/lib/zmap.c
+++ b/lib/zmap.c
@@ -83,6 +83,17 @@ static int z_erofs_fill_inode_lazy(struct erofs_inode *vi)
 		if (ret < 0)
 			return ret;
 	}
+	if (vi->z_advise & Z_EROFS_ADVISE_FRAGMENT_PCLUSTER) {
+		struct erofs_map_blocks map = { .index = UINT_MAX };
+
+		ret = z_erofs_do_map_blocks(vi, &map,
+					    EROFS_GET_BLOCKS_FINDTAIL);
+		if (ret < 0) {
+			erofs_err("failed to find tail for fragment pcluster");
+			return ret;
+		}
+		vi->z_fragmentoff = le32_to_cpu(h->h_fragmentoff);
+	}
 	vi->flags |= EROFS_I_Z_INITED;
 	return 0;
 }
@@ -546,6 +557,7 @@ static int z_erofs_do_map_blocks(struct erofs_inode *vi,
 				 int flags)
 {
 	bool ztailpacking = vi->z_advise & Z_EROFS_ADVISE_INLINE_PCLUSTER;
+	bool in_fragments = vi->z_advise & Z_EROFS_ADVISE_FRAGMENT_PCLUSTER;
 	struct z_erofs_maprecorder m = {
 		.inode = vi,
 		.map = map,
@@ -609,6 +621,8 @@ static int z_erofs_do_map_blocks(struct erofs_inode *vi,
 		map->m_flags |= EROFS_MAP_META;
 		map->m_pa = vi->z_idataoff;
 		map->m_plen = vi->z_idata_size;
+	} else if (in_fragments && m.lcn == vi->z_tailextent_headlcn) {
+		map->m_flags |= EROFS_MAP_FRAGMENTS;
 	} else {
 		map->m_pa = blknr_to_addr(m.pblk);
 		err = z_erofs_get_extent_compressedlen(&m, initial_lcn);
-- 
2.17.1

[RFC PATCH 2/3] erofs-utils: lib: support on-disk offset for shifted decompression

[Yue Hu]

Add support to uncompressed data layout with on-disk offset for
compressed files.

Signed-off-by: Yue Hu <huyue2@coolpad.com>
---
 include/erofs/decompress.h |  3 +++
 lib/data.c                 |  8 +++++++-
 lib/decompress.c           | 10 ++++++++--
 3 files changed, 18 insertions(+), 3 deletions(-)

diff --git a/include/erofs/decompress.h b/include/erofs/decompress.h
index 82bf7b8..b622df5 100644
--- a/include/erofs/decompress.h
+++ b/include/erofs/decompress.h
@@ -23,6 +23,9 @@ struct z_erofs_decompress_req {
 	unsigned int decodedskip;
 	unsigned int inputsize, decodedlength;
 
+	/* head offset of uncompressed data */
+	unsigned int shiftedhead;
+
 	/* indicate the algorithm will be used for decompression */
 	unsigned int alg;
 	bool partial_decoding;
diff --git a/lib/data.c b/lib/data.c
index c2ecbc9..5e44db9 100644
--- a/lib/data.c
+++ b/lib/data.c
@@ -226,7 +226,7 @@ static int z_erofs_read_data(struct erofs_inode *inode, char *buffer,
 	};
 	struct erofs_map_dev mdev;
 	bool partial;
-	unsigned int bufsize = 0;
+	unsigned int bufsize = 0, head;
 	char *raw = NULL;
 	int ret = 0;
 
@@ -313,10 +313,16 @@ static int z_erofs_read_data(struct erofs_inode *inode, char *buffer,
 		if (ret < 0)
 			break;
 
+		head = 0;
+		if (erofs_sb_has_fragments() &&
+		    map.m_algorithmformat == Z_EROFS_COMPRESSION_SHIFTED)
+			head = erofs_blkoff(end);
+
 		ret = z_erofs_decompress(&(struct z_erofs_decompress_req) {
 					.in = raw,
 					.out = buffer + end - offset,
 					.decodedskip = skip,
+					.shiftedhead = head,
 					.inputsize = map.m_plen,
 					.decodedlength = length,
 					.alg = map.m_algorithmformat,
diff --git a/lib/decompress.c b/lib/decompress.c
index 1661f91..08a0861 100644
--- a/lib/decompress.c
+++ b/lib/decompress.c
@@ -132,14 +132,20 @@ out:
 int z_erofs_decompress(struct z_erofs_decompress_req *rq)
 {
 	if (rq->alg == Z_EROFS_COMPRESSION_SHIFTED) {
+		unsigned int count, rightpart;
+
 		if (rq->inputsize > EROFS_BLKSIZ)
 			return -EFSCORRUPTED;
 
 		DBG_BUGON(rq->decodedlength > EROFS_BLKSIZ);
 		DBG_BUGON(rq->decodedlength < rq->decodedskip);
 
-		memcpy(rq->out, rq->in + rq->decodedskip,
-		       rq->decodedlength - rq->decodedskip);
+		count = rq->decodedlength - rq->decodedskip;
+		rightpart = min(EROFS_BLKSIZ - rq->shiftedhead, count);
+
+		memcpy(rq->out, rq->in + (erofs_sb_has_fragments() ?
+		       rq->shiftedhead : rq->decodedskip), rightpart);
+		memcpy(rq->out + rightpart, rq->in, count - rightpart);
 		return 0;
 	}
 
-- 
2.17.1

[RFC PATCH 3/3] erofs-utils: introduce compressed fragments support

[Yue Hu]

This approach can merge tail pclusters or the whole files into a special
inode in order to achieve greater compression ratio. And an option of
pcluster size is provided for different compression requirments.

Meanwhile, we change to write the uncompressed data from 'clusterofs'
when compressing files since it can benefit from in-place I/O. For now,
this change goes with the fragments.

Signed-off-by: Yue Hu <huyue2@coolpad.com>
---
 include/erofs/compress.h  |  2 +-
 include/erofs/config.h    |  3 +-
 include/erofs/fragments.h | 25 +++++++++++
 include/erofs/inode.h     |  2 +
 include/erofs/internal.h  |  1 +
 include/erofs_fs.h        |  5 ++-
 lib/Makefile.am           |  4 +-
 lib/compress.c            | 94 ++++++++++++++++++++++++++++-----------
 lib/fragments.c           | 76 +++++++++++++++++++++++++++++++
 lib/inode.c               | 50 ++++++++++++++-------
 lib/super.c               |  1 +
 mkfs/main.c               | 63 +++++++++++++++++++++++---
 12 files changed, 275 insertions(+), 51 deletions(-)
 create mode 100644 include/erofs/fragments.h
 create mode 100644 lib/fragments.c

diff --git a/include/erofs/compress.h b/include/erofs/compress.h
index 24f6204..fecc316 100644
--- a/include/erofs/compress.h
+++ b/include/erofs/compress.h
@@ -18,7 +18,7 @@ extern "C"
 #define EROFS_CONFIG_COMPR_MIN_SZ           (32   * 1024)
 
 void z_erofs_drop_inline_pcluster(struct erofs_inode *inode);
-int erofs_write_compressed_file(struct erofs_inode *inode);
+int erofs_write_compressed_file_from_fd(struct erofs_inode *inode, int fd);
 
 int z_erofs_compress_init(struct erofs_buffer_head *bh);
 int z_erofs_compress_exit(void);
diff --git a/include/erofs/config.h b/include/erofs/config.h
index 0d0916c..5b83419 100644
--- a/include/erofs/config.h
+++ b/include/erofs/config.h
@@ -44,6 +44,7 @@ struct erofs_configure {
 	char c_chunkbits;
 	bool c_noinline_data;
 	bool c_ztailpacking;
+	bool c_fragments;
 	bool c_ignore_mtime;
 	bool c_showprogress;
 
@@ -62,7 +63,7 @@ struct erofs_configure {
 	/* < 0, xattr disabled and INT_MAX, always use inline xattrs */
 	int c_inline_xattr_tolerance;
 
-	u32 c_pclusterblks_max, c_pclusterblks_def;
+	u32 c_pclusterblks_max, c_pclusterblks_def, c_pclusterblks_frags;
 	u32 c_max_decompressed_extent_bytes;
 	u32 c_dict_size;
 	u64 c_unix_timestamp;
diff --git a/include/erofs/fragments.h b/include/erofs/fragments.h
new file mode 100644
index 0000000..89f0f18
--- /dev/null
+++ b/include/erofs/fragments.h
@@ -0,0 +1,25 @@
+/* SPDX-License-Identifier: GPL-2.0+ OR Apache-2.0 */
+/*
+ * Copyright (C), 2022, Coolpad Group Limited.
+ */
+#ifndef __EROFS_FRAGMENTS_H
+#define __EROFS_FRAGMENTS_H
+
+#ifdef __cplusplus
+extern "C"
+{
+#endif
+
+#include "erofs/internal.h"
+
+int z_erofs_fill_fragments(struct erofs_inode *inode, void *data,
+			   unsigned int len);
+struct erofs_inode *erofs_mkfs_build_fragments(void);
+int erofs_fragments_init(void);
+void erofs_fragments_exit(void);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif
diff --git a/include/erofs/inode.h b/include/erofs/inode.h
index 79b39b0..0a87c58 100644
--- a/include/erofs/inode.h
+++ b/include/erofs/inode.h
@@ -21,6 +21,8 @@ unsigned int erofs_iput(struct erofs_inode *inode);
 erofs_nid_t erofs_lookupnid(struct erofs_inode *inode);
 struct erofs_inode *erofs_mkfs_build_tree_from_path(struct erofs_inode *parent,
 						    const char *path);
+int erofs_prepare_inode_buffer(struct erofs_inode *inode);
+struct erofs_inode *erofs_generate_inode(struct stat64 *st, const char *path);
 
 #ifdef __cplusplus
 }
diff --git a/include/erofs/internal.h b/include/erofs/internal.h
index 129ea54..5a7b2fa 100644
--- a/include/erofs/internal.h
+++ b/include/erofs/internal.h
@@ -193,6 +193,7 @@ struct erofs_inode {
 	unsigned int eof_tailrawsize;
 
 	erofs_off_t fragmentoff;
+	unsigned int fragment_size;
 
 	union {
 		void *compressmeta;
diff --git a/include/erofs_fs.h b/include/erofs_fs.h
index 4a0c74b..4fc2756 100644
--- a/include/erofs_fs.h
+++ b/include/erofs_fs.h
@@ -75,7 +75,9 @@ struct erofs_super_block {
 	} __packed u1;
 	__le16 extra_devices;	/* # of devices besides the primary device */
 	__le16 devt_slotoff;	/* startoff = devt_slotoff * devt_slotsize */
-	__u8 reserved2[38];
+	__u8 reserved[6];
+	__le64 fragments_nid;
+	__u8 reserved2[24];
 };
 
 /*
@@ -265,6 +267,7 @@ struct erofs_inode_chunk_index {
 
 /* maximum supported size of a physical compression cluster */
 #define Z_EROFS_PCLUSTER_MAX_SIZE	(1024 * 1024)
+#define Z_EROFS_PCLUSTER_MAX_BLKS	(Z_EROFS_PCLUSTER_MAX_SIZE / EROFS_BLKSIZ)
 
 /* available compression algorithm types (for h_algorithmtype) */
 enum {
diff --git a/lib/Makefile.am b/lib/Makefile.am
index 3fad357..95f1d55 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -22,12 +22,14 @@ noinst_HEADERS = $(top_srcdir)/include/erofs_fs.h \
       $(top_srcdir)/include/erofs/trace.h \
       $(top_srcdir)/include/erofs/xattr.h \
       $(top_srcdir)/include/erofs/compress_hints.h \
+      $(top_srcdir)/include/erofs/fragments.h \
       $(top_srcdir)/lib/liberofs_private.h
 
 noinst_HEADERS += compressor.h
 liberofs_la_SOURCES = config.c io.c cache.c super.c inode.c xattr.c exclude.c \
 		      namei.c data.c compress.c compressor.c zmap.c decompress.c \
-		      compress_hints.c hashmap.c sha256.c blobchunk.c dir.c
+		      compress_hints.c hashmap.c sha256.c blobchunk.c dir.c \
+		      fragments.c
 liberofs_la_CFLAGS = -Wall -I$(top_srcdir)/include
 if ENABLE_LZ4
 liberofs_la_CFLAGS += ${LZ4_CFLAGS}
diff --git a/lib/compress.c b/lib/compress.c
index ee3b856..145e83e 100644
--- a/lib/compress.c
+++ b/lib/compress.c
@@ -18,6 +18,7 @@
 #include "compressor.h"
 #include "erofs/block_list.h"
 #include "erofs/compress_hints.h"
+#include "erofs/fragments.h"
 
 static struct erofs_compress compresshandle;
 static unsigned int algorithmtype[2];
@@ -35,6 +36,11 @@ struct z_erofs_vle_compress_ctx {
 #define Z_EROFS_LEGACY_MAP_HEADER_SIZE	\
 	(sizeof(struct z_erofs_map_header) + Z_EROFS_VLE_LEGACY_HEADER_PADDING)
 
+static inline bool erofs_has_srcpath(struct erofs_inode *inode)
+{
+	return !(inode->i_srcpath[0] == '\0');
+}
+
 static unsigned int vle_compressmeta_capacity(erofs_off_t filesize)
 {
 	const unsigned int indexsize = BLK_ROUND_UP(filesize) *
@@ -74,9 +80,9 @@ static void vle_write_indexes(struct z_erofs_vle_compress_ctx *ctx,
 	if (!d1) {
 		/*
 		 * A lcluster cannot have three parts with the middle one which
-		 * is well-compressed for !ztailpacking cases.
+		 * is well-compressed for !ztailpacking and !fragments cases.
 		 */
-		DBG_BUGON(!raw && !cfg.c_ztailpacking);
+		DBG_BUGON(!raw && !cfg.c_ztailpacking && !cfg.c_fragments);
 		type = raw ? Z_EROFS_VLE_CLUSTER_TYPE_PLAIN :
 			Z_EROFS_VLE_CLUSTER_TYPE_HEAD;
 		advise = cpu_to_le16(type << Z_EROFS_VLE_DI_CLUSTER_TYPE_BIT);
@@ -143,7 +149,7 @@ static int write_uncompressed_extent(struct z_erofs_vle_compress_ctx *ctx,
 				     unsigned int *len, char *dst)
 {
 	int ret;
-	unsigned int count;
+	unsigned int count, offset, rcopied, rzeroed;
 
 	/* reset clusterofs to 0 if permitted */
 	if (!erofs_sb_has_lz4_0padding() && ctx->clusterofs &&
@@ -153,11 +159,21 @@ static int write_uncompressed_extent(struct z_erofs_vle_compress_ctx *ctx,
 		ctx->clusterofs = 0;
 	}
 
-	/* write uncompressed data */
+	/*
+	 * write uncompressed data from clusterofs which can benefit from
+	 * in-place I/O, loop shift right when to exceed EROFS_BLKSIZ.
+	 */
 	count = min(EROFS_BLKSIZ, *len);
 
-	memcpy(dst, ctx->queue + ctx->head, count);
-	memset(dst + count, 0, EROFS_BLKSIZ - count);
+	offset = cfg.c_fragments ? ctx->clusterofs : 0;
+	rcopied = min(EROFS_BLKSIZ - offset, count);
+	rzeroed = EROFS_BLKSIZ - offset - rcopied;
+
+	memcpy(dst + offset, ctx->queue + ctx->head, rcopied);
+	memcpy(dst, ctx->queue + ctx->head + rcopied, count - rcopied);
+
+	memset(dst + offset + rcopied, 0, rzeroed);
+	memset(dst + count - rcopied, 0, EROFS_BLKSIZ - count - rzeroed);
 
 	erofs_dbg("Writing %u uncompressed data to block %u",
 		  count, ctx->blkaddr);
@@ -169,6 +185,8 @@ static int write_uncompressed_extent(struct z_erofs_vle_compress_ctx *ctx,
 
 static unsigned int z_erofs_get_max_pclusterblks(struct erofs_inode *inode)
 {
+	if (cfg.c_fragments && !erofs_has_srcpath(inode))
+		return cfg.c_pclusterblks_frags;
 #ifndef NDEBUG
 	if (cfg.c_random_pclusterblks)
 		return 1 + rand() % cfg.c_pclusterblks_max;
@@ -237,12 +255,15 @@ static int vle_compress_one(struct erofs_inode *inode,
 		unsigned int pclustersize =
 			z_erofs_get_max_pclusterblks(inode) * EROFS_BLKSIZ;
 		bool may_inline = (cfg.c_ztailpacking && final);
+		bool to_fragments = (cfg.c_fragments && final &&
+				     erofs_has_srcpath(inode));
 		bool raw;
 
 		if (len <= pclustersize) {
 			if (!final)
 				break;
-			if (!may_inline && len <= EROFS_BLKSIZ)
+			if (!may_inline && len <= EROFS_BLKSIZ &&
+			    !to_fragments)
 				goto nocompression;
 		}
 
@@ -294,6 +315,15 @@ nocompression:
 				return ret;
 			ctx->compressedblks = 1;
 			raw = false;
+		} else if (to_fragments && len == count &&
+			   ret < pclustersize) {
+			ret = z_erofs_fill_fragments(inode,
+						     ctx->queue + ctx->head,
+						     len);
+			if (ret < 0)
+				return ret;
+			ctx->compressedblks = 0;
+			raw = false;
 		} else {
 			unsigned int tailused, padding;
 
@@ -546,13 +576,17 @@ static void z_erofs_write_mapheader(struct erofs_inode *inode,
 {
 	struct z_erofs_map_header h = {
 		.h_advise = cpu_to_le16(inode->z_advise),
-		.h_idata_size = cpu_to_le16(inode->idata_size),
 		.h_algorithmtype = inode->z_algorithmtype[1] << 4 |
 				   inode->z_algorithmtype[0],
 		/* lclustersize */
 		.h_clusterbits = inode->z_logical_clusterbits - 12,
 	};
 
+	if (cfg.c_fragments)
+		h.h_fragmentoff = cpu_to_le32(inode->fragmentoff);
+	else
+		h.h_idata_size = cpu_to_le16(inode->idata_size);
+
 	memset(compressmeta, 0, Z_EROFS_LEGACY_MAP_HEADER_SIZE);
 	/* write out map header */
 	memcpy(compressmeta, &h, sizeof(struct z_erofs_map_header));
@@ -604,30 +638,24 @@ void z_erofs_drop_inline_pcluster(struct erofs_inode *inode)
 	inode->eof_tailraw = NULL;
 }
 
-int erofs_write_compressed_file(struct erofs_inode *inode)
+int erofs_write_compressed_file_from_fd(struct erofs_inode *inode, int fd)
 {
 	struct erofs_buffer_head *bh;
 	static struct z_erofs_vle_compress_ctx ctx;
 	erofs_off_t remaining;
 	erofs_blk_t blkaddr, compressed_blocks;
 	unsigned int legacymetasize;
-	int ret, fd;
+	int ret;
 	u8 *compressmeta = malloc(vle_compressmeta_capacity(inode->i_size));
 
 	if (!compressmeta)
 		return -ENOMEM;
 
-	fd = open(inode->i_srcpath, O_RDONLY | O_BINARY);
-	if (fd < 0) {
-		ret = -errno;
-		goto err_free_meta;
-	}
-
 	/* allocate main data buffer */
 	bh = erofs_balloc(DATA, 0, 0, 0);
 	if (IS_ERR(bh)) {
 		ret = PTR_ERR(bh);
-		goto err_close;
+		goto err_free_meta;
 	}
 
 	/* initialize per-file compression setting */
@@ -648,6 +676,9 @@ int erofs_write_compressed_file(struct erofs_inode *inode)
 	inode->z_algorithmtype[1] = algorithmtype[1];
 	inode->z_logical_clusterbits = LOG_BLOCK_SIZE;
 
+	inode->idata_size = 0;
+	inode->fragment_size = 0;
+
 	blkaddr = erofs_mapbh(bh->block);	/* start_blkaddr */
 	ctx.blkaddr = blkaddr;
 	ctx.metacur = compressmeta + Z_EROFS_LEGACY_MAP_HEADER_SIZE;
@@ -681,19 +712,20 @@ int erofs_write_compressed_file(struct erofs_inode *inode)
 	vle_write_indexes_final(&ctx);
 	legacymetasize = ctx.metacur - compressmeta;
 	/* estimate if data compression saves space or not */
-	if (compressed_blocks * EROFS_BLKSIZ + inode->idata_size +
+	if (!inode->fragment_size &&
+	    compressed_blocks * EROFS_BLKSIZ + inode->idata_size +
 	    legacymetasize >= inode->i_size) {
 		ret = -ENOSPC;
 		goto err_free_idata;
 	}
 	z_erofs_write_mapheader(inode, compressmeta);
 
-	close(fd);
 	if (compressed_blocks) {
 		ret = erofs_bh_balloon(bh, blknr_to_addr(compressed_blocks));
 		DBG_BUGON(ret != EROFS_BLKSIZ);
 	} else {
-		DBG_BUGON(!inode->idata_size);
+		if (!cfg.c_fragments)
+			DBG_BUGON(!inode->idata_size);
 	}
 
 	erofs_info("compressed %s (%llu bytes) into %u blocks",
@@ -716,7 +748,8 @@ int erofs_write_compressed_file(struct erofs_inode *inode)
 		DBG_BUGON(ret);
 	}
 	inode->compressmeta = compressmeta;
-	erofs_droid_blocklist_write(inode, blkaddr, compressed_blocks);
+	if (erofs_has_srcpath(inode))
+		erofs_droid_blocklist_write(inode, blkaddr, compressed_blocks);
 	return 0;
 
 err_free_idata:
@@ -726,8 +759,6 @@ err_free_idata:
 	}
 err_bdrop:
 	erofs_bdrop(bh, true);	/* revoke buffer */
-err_close:
-	close(fd);
 err_free_meta:
 	free(compressmeta);
 	return ret;
@@ -833,14 +864,27 @@ int z_erofs_compress_init(struct erofs_buffer_head *sb_bh)
 	 * to be loaded in order to get those compressed block counts.
 	 */
 	if (cfg.c_pclusterblks_max > 1) {
-		if (cfg.c_pclusterblks_max >
-		    Z_EROFS_PCLUSTER_MAX_SIZE / EROFS_BLKSIZ) {
+		if (cfg.c_pclusterblks_max > Z_EROFS_PCLUSTER_MAX_BLKS) {
 			erofs_err("unsupported clusterblks %u (too large)",
 				  cfg.c_pclusterblks_max);
 			return -EINVAL;
 		}
+		if (cfg.c_pclusterblks_frags > Z_EROFS_PCLUSTER_MAX_BLKS) {
+			erofs_err("unsupported clusterblks %u (too large for fragments)",
+				  cfg.c_pclusterblks_frags);
+			return -EINVAL;
+		}
+		if (cfg.c_pclusterblks_frags == 1) {
+			erofs_err("physical cluster size of fragments should > 4096 bytes");
+			return -EINVAL;
+		}
 		erofs_sb_set_big_pcluster();
 	}
+	if (!erofs_sb_has_big_pcluster() && cfg.c_pclusterblks_frags > 1) {
+		erofs_err("invalid clusterblks %u (for fragments)",
+			  cfg.c_pclusterblks_frags);
+		return -EINVAL;
+	}
 
 	if (ret != Z_EROFS_COMPRESSION_LZ4)
 		erofs_sb_set_compr_cfgs();
diff --git a/lib/fragments.c b/lib/fragments.c
new file mode 100644
index 0000000..e74eec2
--- /dev/null
+++ b/lib/fragments.c
@@ -0,0 +1,76 @@
+// SPDX-License-Identifier: GPL-2.0+ OR Apache-2.0
+/*
+ * Copyright (C), 2022, Coolpad Group Limited.
+ * Created by Yue Hu <huyue2@coolpad.com>
+ */
+#define _GNU_SOURCE
+#include <stdlib.h>
+#include <unistd.h>
+#include <sys/stat.h>
+#include "erofs/err.h"
+#include "erofs/inode.h"
+#include "erofs/compress.h"
+#include "erofs/print.h"
+#include "erofs/fragments.h"
+
+static FILE *fragmentsfp;
+
+int z_erofs_fill_fragments(struct erofs_inode *inode, void *data,
+			   unsigned int len)
+{
+	inode->z_advise |= Z_EROFS_ADVISE_FRAGMENT_PCLUSTER;
+	inode->fragmentoff = ftell(fragmentsfp);
+	inode->fragment_size = len;
+
+	if (write(fileno(fragmentsfp), data, len) < 0)
+		return -EIO;
+
+	erofs_sb_set_fragments();
+
+	erofs_dbg("Recording %u fragment data at %lu", inode->fragment_size,
+		  inode->fragmentoff);
+	return len;
+}
+
+struct erofs_inode *erofs_mkfs_build_fragments(void)
+{
+	struct stat64 st;
+	struct erofs_inode *inode;
+	int ret, fd = fileno(fragmentsfp);
+
+	ret = fstat64(fd, &st);
+	if (ret)
+		return ERR_PTR(-errno);
+
+	inode = erofs_generate_inode(&st, NULL);
+	if (IS_ERR(inode))
+		return inode;
+
+	fseek(fragmentsfp, 0, SEEK_SET);
+	ret = erofs_write_compressed_file_from_fd(inode, fd);
+	if (ret) {
+		erofs_err("write fragments file error");
+		return ERR_PTR(ret);
+	}
+
+	erofs_prepare_inode_buffer(inode);
+	return inode;
+}
+
+void erofs_fragments_exit(void)
+{
+	if (fragmentsfp)
+		fclose(fragmentsfp);
+}
+
+int erofs_fragments_init(void)
+{
+#ifdef HAVE_TMPFILE64
+	fragmentsfp = tmpfile64();
+#else
+	fragmentsfp = tmpfile();
+#endif
+	if (!fragmentsfp)
+		return -ENOMEM;
+	return 0;
+}
diff --git a/lib/inode.c b/lib/inode.c
index f192510..86adbc4 100644
--- a/lib/inode.c
+++ b/lib/inode.c
@@ -405,7 +405,11 @@ int erofs_write_file(struct erofs_inode *inode)
 	}
 
 	if (cfg.c_compr_alg_master && erofs_file_is_compressible(inode)) {
-		ret = erofs_write_compressed_file(inode);
+		fd = open(inode->i_srcpath, O_RDONLY | O_BINARY);
+		if (fd < 0)
+			return -errno;
+		ret = erofs_write_compressed_file_from_fd(inode, fd);
+		close(fd);
 
 		if (!ret || ret != -ENOSPC)
 			return ret;
@@ -583,7 +587,7 @@ static int erofs_prepare_tail_block(struct erofs_inode *inode)
 	return 0;
 }
 
-static int erofs_prepare_inode_buffer(struct erofs_inode *inode)
+int erofs_prepare_inode_buffer(struct erofs_inode *inode)
 {
 	unsigned int inodesize;
 	struct erofs_buffer_head *bh, *ibh;
@@ -782,6 +786,9 @@ int erofs_droid_inode_fsconfig(struct erofs_inode *inode,
 	const char *fspath;
 	char *decorated = NULL;
 
+	if (!path)
+		return 0;
+
 	inode->capabilities = 0;
 	if (!cfg.fs_config_file && !cfg.mount_point)
 		return 0;
@@ -868,14 +875,18 @@ static int erofs_fill_inode(struct erofs_inode *inode,
 		return -EINVAL;
 	}
 
-	strncpy(inode->i_srcpath, path, sizeof(inode->i_srcpath) - 1);
-	inode->i_srcpath[sizeof(inode->i_srcpath) - 1] = '\0';
+	if (path) {
+		strncpy(inode->i_srcpath, path, sizeof(inode->i_srcpath) - 1);
+		inode->i_srcpath[sizeof(inode->i_srcpath) - 1] = '\0';
+	} else {
+		inode->i_srcpath[0] = '\0';
+	}
 
 	inode->dev = st->st_dev;
 	inode->i_ino[1] = st->st_ino;
 
 	if (erofs_should_use_inode_extended(inode)) {
-		if (cfg.c_force_inodeversion == FORCE_INODE_COMPACT) {
+		if (path && cfg.c_force_inodeversion == FORCE_INODE_COMPACT) {
 			erofs_err("file %s cannot be in compact form",
 				  inode->i_srcpath);
 			return -EINVAL;
@@ -907,6 +918,23 @@ static struct erofs_inode *erofs_new_inode(void)
 	return inode;
 }
 
+struct erofs_inode *erofs_generate_inode(struct stat64 *st, const char *path)
+{
+	struct erofs_inode *inode;
+	int ret;
+
+	inode = erofs_new_inode();
+	if (IS_ERR(inode))
+		return inode;
+
+	ret = erofs_fill_inode(inode, st, path);
+	if (ret) {
+		free(inode);
+		return ERR_PTR(ret);
+	}
+	return inode;
+}
+
 /* get the inode from the (source) path */
 static struct erofs_inode *erofs_iget_from_path(const char *path, bool is_src)
 {
@@ -934,17 +962,7 @@ static struct erofs_inode *erofs_iget_from_path(const char *path, bool is_src)
 	}
 
 	/* cannot find in the inode cache */
-	inode = erofs_new_inode();
-	if (IS_ERR(inode))
-		return inode;
-
-	ret = erofs_fill_inode(inode, &st, path);
-	if (ret) {
-		free(inode);
-		return ERR_PTR(ret);
-	}
-
-	return inode;
+	return erofs_generate_inode(&st, path);
 }
 
 static void erofs_fixup_meta_blkaddr(struct erofs_inode *rootdir)
diff --git a/lib/super.c b/lib/super.c
index f486eb7..e1cf614 100644
--- a/lib/super.c
+++ b/lib/super.c
@@ -100,6 +100,7 @@ int erofs_read_superblock(void)
 	sbi.xattr_blkaddr = le32_to_cpu(dsb->xattr_blkaddr);
 	sbi.islotbits = EROFS_ISLOTBITS;
 	sbi.root_nid = le16_to_cpu(dsb->root_nid);
+	sbi.fragments_nid = le64_to_cpu(dsb->fragments_nid);
 	sbi.inos = le64_to_cpu(dsb->inos);
 	sbi.checksum = le32_to_cpu(dsb->checksum);
 
diff --git a/mkfs/main.c b/mkfs/main.c
index d2c9830..227a9be 100644
--- a/mkfs/main.c
+++ b/mkfs/main.c
@@ -23,6 +23,7 @@
 #include "erofs/block_list.h"
 #include "erofs/compress_hints.h"
 #include "erofs/blobchunk.h"
+#include "erofs/fragments.h"
 #include "../lib/liberofs_private.h"
 
 #ifdef HAVE_LIBUUID
@@ -129,9 +130,9 @@ static int parse_extended_opts(const char *opts)
 		const char *p = strchr(token, ',');
 
 		next = NULL;
-		if (p)
+		if (p) {
 			next = p + 1;
-		else {
+		} else {
 			p = token + strlen(token);
 			next = p;
 		}
@@ -198,7 +199,34 @@ static int parse_extended_opts(const char *opts)
 				return -EINVAL;
 			cfg.c_ztailpacking = true;
 		}
+
+		if (MATCH_EXTENTED_OPT("fragments", token, keylen)) {
+			char *endptr;
+			u64 i;
+
+			if (vallen || cfg.c_ztailpacking)
+				return -EINVAL;
+			cfg.c_fragments = true;
+
+			i = strtoull(next, &endptr, 0);
+			if (i == 0 || (*endptr != ',' && *endptr != '\0')) {
+				cfg.c_pclusterblks_frags = 1;
+				continue;
+			}
+			if (i % EROFS_BLKSIZ) {
+				erofs_err("invalid physical clustersize %llu",
+					  i);
+				return -EINVAL;
+			}
+			cfg.c_pclusterblks_frags = i / EROFS_BLKSIZ;
+
+			if (*endptr == ',')
+				next = strchr(next, ',')  + 1;
+			else
+				goto out;
+		}
 	}
+out:
 	return 0;
 }
 
@@ -438,7 +466,8 @@ static int mkfs_parse_options_cfg(int argc, char *argv[])
 
 int erofs_mkfs_update_super_block(struct erofs_buffer_head *bh,
 				  erofs_nid_t root_nid,
-				  erofs_blk_t *blocks)
+				  erofs_blk_t *blocks,
+				  erofs_nid_t fragments_nid)
 {
 	struct erofs_super_block sb = {
 		.magic     = cpu_to_le32(EROFS_SUPER_MAGIC_V1),
@@ -462,6 +491,7 @@ int erofs_mkfs_update_super_block(struct erofs_buffer_head *bh,
 	*blocks         = erofs_mapbh(NULL);
 	sb.blocks       = cpu_to_le32(*blocks);
 	sb.root_nid     = cpu_to_le16(root_nid);
+	sb.fragments_nid     = cpu_to_le64(fragments_nid);
 	memcpy(sb.uuid, sbi.uuid, sizeof(sb.uuid));
 
 	if (erofs_sb_has_compr_cfgs())
@@ -579,8 +609,8 @@ int main(int argc, char **argv)
 {
 	int err = 0;
 	struct erofs_buffer_head *sb_bh;
-	struct erofs_inode *root_inode;
-	erofs_nid_t root_nid;
+	struct erofs_inode *root_inode, *fragments_inode;
+	erofs_nid_t root_nid, fragments_nid;
 	struct stat64 st;
 	erofs_blk_t nblocks;
 	struct timeval t;
@@ -650,6 +680,14 @@ int main(int argc, char **argv)
 		erofs_warn("EXPERIMENTAL chunked file feature in use. Use at your own risk!");
 	if (cfg.c_ztailpacking)
 		erofs_warn("EXPERIMENTAL compressed inline data feature in use. Use at your own risk!");
+	if (cfg.c_fragments) {
+		err = erofs_fragments_init();
+		if (err) {
+			erofs_err("failed to initialize fragments");
+			return 1;
+		}
+		erofs_warn("EXPERIMENTAL compressed fragments feature in use. Use at your own risk!");
+	}
 	erofs_set_fs_root(cfg.c_src_path);
 #ifndef NDEBUG
 	if (cfg.c_random_pclusterblks)
@@ -719,7 +757,18 @@ int main(int argc, char **argv)
 			goto exit;
 	}
 
-	err = erofs_mkfs_update_super_block(sb_bh, root_nid, &nblocks);
+	fragments_nid = 0;
+	if (cfg.c_fragments) {
+		fragments_inode = erofs_mkfs_build_fragments();
+		if (IS_ERR(fragments_inode)) {
+			err = PTR_ERR(fragments_inode);
+			goto exit;
+		}
+		fragments_nid = erofs_lookupnid(fragments_inode);
+	}
+
+	err = erofs_mkfs_update_super_block(sb_bh, root_nid, &nblocks,
+					    fragments_nid);
 	if (err)
 		goto exit;
 
@@ -741,6 +790,8 @@ exit:
 	erofs_cleanup_exclude_rules();
 	if (cfg.c_chunkbits)
 		erofs_blob_exit();
+	if (cfg.c_fragments)
+		erofs_fragments_exit();
 	erofs_exit_configure();
 
 	if (err) {
-- 
2.17.1

Re: [RFC PATCH 0/3] erofs-utils: compressed fragments feature

[Gao Xiang]

Hi Yue,

On Mon, Jul 11, 2022 at 05:09:55PM +0800, Yue Hu wrote:
> In order to achieve greater compression ratio, let's introduce
> compressed fragments feature which can merge tail of per-file or the
> whole files into one special inode to reach the target.
> 
> And we can also set pcluster size to fragments inode for different
> compression requirments.
> 
> In this patchset, we also improve the uncompressed data layout of
> compressed files. Just write it from 'clusterofs' instead of 0 since it
> can benefit from in-place I/O. For now, it only goes with fragments.
> 
> The main idea above is from Xiang.
> 

I just took a preliminary try and it seems work, but in order to form it
better, I have to postpone it for the next version of 5.20 (6.0?)
since I'm still working on cleaning up and rolling hash deduplication.

Thanks,
Gao Xiang

Re: [RFC PATCH 0/3] erofs-utils: compressed fragments feature

[Yue Hu]

Hi Xiang,

On Mon, 11 Jul 2022 18:02:43 +0800
Gao Xiang <hsiangkao@linux.alibaba.com> wrote:

> Hi Yue,
> 
> On Mon, Jul 11, 2022 at 05:09:55PM +0800, Yue Hu wrote:
> > In order to achieve greater compression ratio, let's introduce
> > compressed fragments feature which can merge tail of per-file or the
> > whole files into one special inode to reach the target.
> > 
> > And we can also set pcluster size to fragments inode for different
> > compression requirments.
> > 
> > In this patchset, we also improve the uncompressed data layout of
> > compressed files. Just write it from 'clusterofs' instead of 0 since it
> > can benefit from in-place I/O. For now, it only goes with fragments.
> > 
> > The main idea above is from Xiang.
> >   
> 
> I just took a preliminary try and it seems work, but in order to form it
> better, I have to postpone it for the next version of 5.20 (6.0?)
> since I'm still working on cleaning up and rolling hash deduplication.

Got it.

Thanks.

> 
> Thanks,
> Gao Xiang