1 ===============================================================================
3 ===============================================================================
5 exofs is a file system that uses an OSD and exports the API of a normal Linux
6 file system. Users access exofs like any other local file system, and exofs
7 will in turn issue commands to the local OSD initiator.
9 OSD is a new T10 command set that views storage devices not as a large/flat
10 array of sectors but as a container of objects, each having a length, quota,
11 time attributes and more. Each object is addressed by a 64bit ID, and is
12 contained in a 64bit ID partition. Each object has associated attributes
13 attached to it, which are integral part of the object and provide metadata about
14 the object. The standard defines some common obligatory attributes, but user
15 attributes can be added as needed.
17 ===============================================================================
19 ===============================================================================
21 To use this file system, you need to have an object store to run it on. You
22 may download a target from:
25 See Documentation/scsi/osd.txt for how to setup a working osd environment.
27 ===============================================================================
29 ===============================================================================
31 1. Download and compile exofs and open-osd initiator:
32 You need an external Kernel source tree or kernel headers from your
33 distribution. (anything based on 2.6.26 or later).
35 a. download open-osd including exofs source using:
36 [parent-directory]$ git clone git://git.open-osd.org/open-osd.git
38 b. Build the library module like this:
39 [parent-directory]$ make -C KSRC=$(KER_DIR) open-osd
41 This will build both the open-osd initiator as well as the exofs kernel
42 module. Use whatever parameters you compiled your Kernel with and
43 $(KER_DIR) above pointing to the Kernel you compile against. See the file
44 open-osd/top-level-Makefile for an example.
46 2. Get the OSD initiator and target set up properly, and login to the target.
47 See Documentation/scsi/osd.txt for farther instructions. Also see ./do-osd
48 for example script that does all these steps.
50 3. Insmod the exofs.ko module:
51 [exofs]$ insmod exofs.ko
53 4. Make sure the directory where you want to mount exists. If not, create it.
54 (For example, mkdir /mnt/exofs)
56 5. At first run you will need to invoke the mkfs.exofs application
58 As an example, this will create the file system on:
59 /dev/osd0 partition ID 65536
61 mkfs.exofs --pid=65536 --format /dev/osd0
63 The --format is optional if not specified no OSD_FORMAT will be
64 preformed and a clean file system will be created in the specified pid,
65 in the available space of the target. (Use --format=size_in_meg to limit
66 the total LUN space available)
68 If pid already exist it will be deleted and a new one will be created in it's
71 An exofs lives inside a single OSD partition. You can create multiple exofs
72 filesystems on the same device using multiple pids.
74 (run mkfs.exofs without any parameters for usage help message)
76 6. Mount the file system.
78 For example, to mount /dev/osd0, partition ID 0x10000 on /mnt/exofs:
80 mount -t exofs -o pid=65536 /dev/osd0 /mnt/exofs/
82 7. For reference (See do-exofs example script):
83 do-exofs start - an example of how to perform the above steps.
84 do-exofs stop - an example of how to unmount the file system.
85 do-exofs format - an example of how to format and mkfs a new exofs.
87 8. Extra compilation flags (uncomment in fs/exofs/Kbuild):
88 CONFIG_EXOFS_DEBUG - for debug messages and extra checks.
90 ===============================================================================
92 ===============================================================================
93 Similar to any mount command:
94 mount -t exofs -o exofs_options /dev/osdX mount_exofs_directory
97 -t exofs: specifies the exofs file system
99 /dev/osdX: X is a decimal number. /dev/osdX was created after a successful
100 login into an OSD target.
102 mount_exofs_directory: The directory to mount the file system on
104 exofs specific options: Options are separated by commas (,)
105 pid=<integer> - The partition number to mount/create as
106 container of the filesystem.
107 This option is mandatory
108 to=<integer> - Timeout in ticks for a single command
109 default is (60 * HZ) [for debugging only]
111 ===============================================================================
113 ===============================================================================
115 * The file system control block (AKA on-disk superblock) resides in an object
116 with a special ID (defined in common.h).
117 Information included in the file system control block is used to fill the
118 in-memory superblock structure at mount time. This object is created before
119 the file system is used by mkexofs.c It contains information such as:
120 - The file system's magic number
121 - The next inode number to be allocated
123 * Each file resides in its own object and contains the data (and it will be
124 possible to extend the file over multiple objects, though this has not been
127 * A directory is treated as a file, and essentially contains a list of <file
128 name, inode #> pairs for files that are found in that directory. The object
129 IDs correspond to the files' inode numbers and will be allocated according to
130 a bitmap (stored in a separate object). Now they are allocated using a
133 * Each file's control block (AKA on-disk inode) is stored in its object's
134 attributes. This applies to both regular files and other types (directories,
135 device files, symlinks, etc.).
137 * Credentials are generated per object (inode and superblock) when they is
138 created in memory (read off disk or created). The credential works for all
139 operations and is used as long as the object remains in memory.
141 * Async OSD operations are used whenever possible, but the target may execute
142 them out of order. The operations that concern us are create, delete,
143 readpage, writepage, update_inode, and truncate. The following pairs of
144 operations should execute in the order written, and we need to prevent them
145 from executing in reverse order:
146 - The following are handled with the OBJ_CREATED and OBJ_2BCREATED
147 flags. OBJ_CREATED is set when we know the object exists on the OSD -
148 in create's callback function, and when we successfully do a read_inode.
149 OBJ_2BCREATED is set in the beginning of the create function, so we
150 know that we should wait.
151 - create/delete: delete should wait until the object is created
153 - create/readpage: readpage should be able to return a page
154 full of zeroes in this case. If there was a write already
155 en-route (i.e. create, writepage, readpage) then the page
156 would be locked, and so it would really be the same as
158 - create/writepage: if writepage is called for a sync write, it
159 should wait until the object is created on the OSD.
160 Otherwise, it should just return.
161 - create/truncate: truncate should wait until the object is
163 - create/update_inode: update_inode should wait until the
164 object is created on the OSD.
165 - Handled by VFS locks:
166 - readpage/delete: shouldn't happen because of page lock.
167 - writepage/delete: shouldn't happen because of page lock.
168 - readpage/writepage: shouldn't happen because of page lock.
170 ===============================================================================
172 ===============================================================================
173 The exofs file system is based on ext2 v0.5b (distributed with the Linux kernel
174 version 2.6.10). All files include the original copyrights, and the license
175 is GPL version 2 (only version 2, as is true for the Linux kernel). The
176 Linux kernel can be downloaded from www.kernel.org.