1 .. SPDX-License-Identifier: GPL-2.0
3 ======================================
4 Enhanced Read-Only File System - EROFS
5 ======================================
10 EROFS file-system stands for Enhanced Read-Only File System. Different
11 from other read-only file systems, it aims to be designed for flexibility,
12 scalability, but be kept simple and high performance.
14 It is designed as a better filesystem solution for the following scenarios:
16 - read-only storage media or
18 - part of a fully trusted read-only solution, which means it needs to be
19 immutable and bit-for-bit identical to the official golden image for
20 their releases due to security and other considerations and
22 - hope to save some extra storage space with guaranteed end-to-end performance
23 by using reduced metadata and transparent file compression, especially
24 for those embedded devices with limited memory (ex, smartphone);
26 Here is the main features of EROFS:
28 - Little endian on-disk design;
30 - Currently 4KB block size (nobh) and therefore maximum 16TB address space;
32 - Metadata & data could be mixed by design;
34 - 2 inode versions for different requirements:
36 ===================== ============ =====================================
37 compact (v1) extended (v2)
38 ===================== ============ =====================================
39 Inode metadata size 32 bytes 64 bytes
40 Max file size 4 GB 16 EB (also limited by max. vol size)
41 Max uids/gids 65536 4294967296
42 File change time no yes (64 + 32-bit timestamp)
43 Max hardlinks 65536 4294967296
44 Metadata reserved 4 bytes 14 bytes
45 ===================== ============ =====================================
47 - Support extended attributes (xattrs) as an option;
49 - Support xattr inline and tail-end data inline for all files;
51 - Support POSIX.1e ACLs by using xattrs;
53 - Support transparent data compression as an option:
54 LZ4 algorithm with the fixed-sized output compression for high performance.
56 The following git tree provides the file system user-space tools under
57 development (ex, formatting tool mkfs.erofs):
59 - git://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs-utils.git
61 Bugs and patches are welcome, please kindly help us and send to the following
62 linux-erofs mailing list:
64 - linux-erofs mailing list <linux-erofs@lists.ozlabs.org>
69 =================== =========================================================
70 (no)user_xattr Setup Extended User Attributes. Note: xattr is enabled
71 by default if CONFIG_EROFS_FS_XATTR is selected.
72 (no)acl Setup POSIX Access Control List. Note: acl is enabled
73 by default if CONFIG_EROFS_FS_POSIX_ACL is selected.
74 cache_strategy=%s Select a strategy for cached decompression from now on:
76 ========== =============================================
77 disabled In-place I/O decompression only;
78 readahead Cache the last incomplete compressed physical
79 cluster for further reading. It still does
80 in-place I/O decompression for the rest
81 compressed physical clusters;
82 readaround Cache the both ends of incomplete compressed
83 physical clusters for further reading.
84 It still does in-place I/O decompression
85 for the rest compressed physical clusters.
86 ========== =============================================
87 =================== =========================================================
94 Different from other read-only file systems, an EROFS volume is designed
95 to be as simple as possible::
97 |-> aligned with the block size
98 ____________________________________________________________
99 | |SB| | ... | Metadata | ... | Data | Metadata | ... | Data |
100 |_|__|_|_____|__________|_____|______|__________|_____|______|
103 All data areas should be aligned with the block size, but metadata areas
104 may not. All metadatas can be now observed in two different spaces (views):
106 1. Inode metadata space
108 Each valid inode should be aligned with an inode slot, which is a fixed
109 value (32 bytes) and designed to be kept in line with compact inode size.
111 Each inode can be directly found with the following formula:
112 inode offset = meta_blkaddr * block_size + 32 * nid
118 + meta_blkaddr blocks |-> another slot
119 _____________________________________________________________________
120 | ... | inode | xattrs | extents | data inline | ... | inode ...
121 |________|_______|(optional)|(optional)|__(optional)_|_____|__________
122 |-> aligned with the inode slot size
129 .____________________________________________________|-> aligned with 4B
130 | xattr_ibody_header | shared xattrs | inline xattrs |
131 |____________________|_______________|_______________|
132 |-> 12 bytes <-|->x * 4 bytes<-| .
136 ._______________________________.______________________.
137 | id | id | id | id | ... | id | ent | ... | ent| ... |
138 |____|____|____|____|______|____|_____|_____|____|_____|
142 Inode could be 32 or 64 bytes, which can be distinguished from a common
143 field which all inode versions have -- i_format::
145 __________________ __________________
146 | i_format | | i_format |
147 |__________________| |__________________|
150 |__________________| 32 bytes | |
152 |__________________| 64 bytes
154 Xattrs, extents, data inline are followed by the corresponding inode with
155 proper alignment, and they could be optional for different data mappings.
156 _currently_ total 4 valid data mappings are supported:
158 == ====================================================================
159 0 flat file data without data inline (no extent);
160 1 fixed-sized output data compression (with non-compacted indexes);
161 2 flat file data with tail packing data inline (no extent);
162 3 fixed-sized output data compression (with compacted indexes, v5.3+).
163 == ====================================================================
165 The size of the optional xattrs is indicated by i_xattr_count in inode
166 header. Large xattrs or xattrs shared by many different files can be
167 stored in shared xattrs metadata rather than inlined right after inode.
169 2. Shared xattrs metadata space
171 Shared xattrs space is similar to the above inode space, started with
172 a specific block indicated by xattr_blkaddr, organized one by one with
175 Each share xattr can also be directly found by the following formula:
176 xattr offset = xattr_blkaddr * block_size + 4 * xattr_id
180 |-> aligned by 4 bytes
181 + xattr_blkaddr blocks |-> aligned with 4 bytes
182 _________________________________________________________________________
183 | ... | xattr_entry | xattr data | ... | xattr_entry | xattr data ...
184 |________|_____________|_____________|_____|______________|_______________
188 All directories are now organized in a compact on-disk format. Note that
189 each directory block is divided into index and name areas in order to support
190 random file lookup, and all directory entries are _strictly_ recorded in
191 alphabetical order in order to support improved prefix binary search
192 algorithm (could refer to the related source code).
196 ___________________________
198 / ______________|________________
199 / / | nameoff1 | nameoffN-1
200 ____________.______________._______________v________________v__________
201 | dirent | dirent | ... | dirent | filename | filename | ... | filename |
202 |___.0___|____1___|_____|___N-1__|____0_____|____1_____|_____|___N-1____|
206 \________________________| nameoff0
209 Note that apart from the offset of the first filename, nameoff0 also indicates
210 the total number of directory entries in this block since it is no need to
211 introduce another on-disk field at all.
215 EROFS implements LZ4 fixed-sized output compression which generates fixed-sized
216 compressed data blocks from variable-sized input in contrast to other existing
217 fixed-sized input solutions. Relatively higher compression ratios can be gotten
218 by using fixed-sized output compression since nowadays popular data compression
219 algorithms are mostly LZ77-based and such fixed-sized output approach can be
220 benefited from the historical dictionary (aka. sliding window).
222 In details, original (uncompressed) data is turned into several variable-sized
223 extents and in the meanwhile, compressed into physical clusters (pclusters).
224 In order to record each variable-sized extent, logical clusters (lclusters) are
225 introduced as the basic unit of compress indexes to indicate whether a new
226 extent is generated within the range (HEAD) or not (NONHEAD). Lclusters are now
227 fixed in block size, as illustrated below::
229 |<- variable-sized extent ->|<- VLE ->|
230 clusterofs clusterofs clusterofs
232 _________v_________________________________v_______________________v________
233 ... | . | | . | | . ...
234 ____|____._________|______________|________.___ _|______________|__.________
235 |-> lcluster <-|-> lcluster <-|-> lcluster <-|-> lcluster <-|
236 (HEAD) (NONHEAD) (HEAD) (NONHEAD) .
240 _______._____________________________.______________._________________
242 _______|______________|______________|______________|_________________
243 |-> big pcluster <-|-> pcluster <-|
245 A physical cluster can be seen as a container of physical compressed blocks
246 which contains compressed data. Previously, only lcluster-sized (4KB) pclusters
247 were supported. After big pcluster feature is introduced (available since
248 Linux v5.13), pcluster can be a multiple of lcluster size.
250 For each HEAD lcluster, clusterofs is recorded to indicate where a new extent
251 starts and blkaddr is used to seek the compressed data. For each NONHEAD
252 lcluster, delta0 and delta1 are available instead of blkaddr to indicate the
253 distance to its HEAD lcluster and the next HEAD lcluster. A PLAIN lcluster is
254 also a HEAD lcluster except that its data is uncompressed. See the comments
255 around "struct z_erofs_vle_decompressed_index" in erofs_fs.h for more details.
257 If big pcluster is enabled, pcluster size in lclusters needs to be recorded as
258 well. Let the delta0 of the first NONHEAD lcluster store the compressed block
259 count with a special flag as a new called CBLKCNT NONHEAD lcluster. It's easy
260 to understand its delta0 is constantly 1, as illustrated below::
262 __________________________________________________________
263 | HEAD | NONHEAD | NONHEAD | ... | NONHEAD | HEAD | HEAD |
264 |__:___|_(CBLKCNT)_|_________|_____|_________|__:___|____:_|
265 |<----- a big pcluster (with CBLKCNT) ------>|<-- -->|
266 a lcluster-sized pcluster (without CBLKCNT) ^
268 If another HEAD follows a HEAD lcluster, there is no room to record CBLKCNT,
269 but it's easy to know the size of such pcluster is 1 lcluster as well.