Skip to contentSkip to conversation

erofs-utils: compressed fragments feature

RFC PATCH Superseded Yue Hu 2 participants 6 messages Jul 11, 2022 Library code

v1 of 6 Superseded Latest is v6 →

lore
Share & tools
lore

Yue Hu

author
#0 lore raw

cover erofs-utils: compressed fragments feature

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
Signature · 1 line
2.17.1
author
#1 lore raw

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

Add compressed fragments support for erofsfuse.

---
 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(-)
include/erofs/internal.h +8 -0
index 6a70f11..129ea54 100644
@@ -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];
include/erofs_fs.h +11 -4
index 08f9761..4a0c74b 100644
@@ -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);
lib/data.c +26 -0
index 6bc554d..c2ecbc9 100644
@@ -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);
lib/zmap.c +14 -0
index 95745c5..2b45a4f 100644
@@ -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);
Signature · 1 line
2.17.1
author
#2 lore raw

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

In reply to Yue Hu

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

---
 include/erofs/decompress.h |  3 +++
 lib/data.c                 |  8 +++++++-
 lib/decompress.c           | 10 ++++++++--
 3 files changed, 18 insertions(+), 3 deletions(-)
include/erofs/decompress.h +3 -0
index 82bf7b8..b622df5 100644
@@ -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;
lib/data.c +7 -1
index c2ecbc9..5e44db9 100644
@@ -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,
lib/decompress.c +8 -2
index 1661f91..08a0861 100644
@@ -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;
 	}
 
Signature · 1 line
2.17.1
author
#3 lore raw

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

In reply to 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.

---
 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
include/erofs/compress.h +1 -1
index 24f6204..fecc316 100644
@@ -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);
include/erofs/config.h +2 -1
index 0d0916c..5b83419 100644
@@ -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;
include/erofs/fragments.h new file +25 -0
new file mode 100644
index 0000000..89f0f18
@@ -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
include/erofs/inode.h +2 -0
index 79b39b0..0a87c58 100644
@@ -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
 }
include/erofs/internal.h +1 -0
index 129ea54..5a7b2fa 100644
@@ -193,6 +193,7 @@ struct erofs_inode {
 	unsigned int eof_tailrawsize;
 
 	erofs_off_t fragmentoff;
+	unsigned int fragment_size;
 
 	union {
 		void *compressmeta;
include/erofs_fs.h +4 -1
index 4a0c74b..4fc2756 100644
@@ -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 {
lib/Makefile.am +3 -1
index 3fad357..95f1d55 100644
@@ -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}
lib/compress.c +69 -25

229 lines not on this page · raw message

lib/fragments.c new file +76 -0

76 lines not on this page · raw message

lib/inode.c +34 -16

91 lines not on this page · raw message

lib/super.c +1 -0
index f486eb7..e1cf614 100644
@@ -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);
 
mkfs/main.c +57 -6

119 lines not on this page · raw message

Signature · 1 line
2.17.1
#4 lore raw
In reply to Yue Hu

Hi Yue,

Quoted 4 lines · 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.

Quoted 9 lines

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

author
#5 lore raw

Hi Xiang,

Quoted 14 lines · On Mon, 11 Jul 2022 18:02:43 +0800 Gao Xiang <[email protected]>…

Hi Yue,

Quoted 11 lines · 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.

Quoted 6 lines
Quoted 2 lines

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.

Quoted 3 lines

Thanks,
Gao Xiang