xref: /linux/Documentation/filesystems/9p.rst (revision 3a39d672e7f48b8d6b91a09afa4b55352773b4b5)
1.. SPDX-License-Identifier: GPL-2.0
2
3=======================================
4v9fs: Plan 9 Resource Sharing for Linux
5=======================================
6
7About
8=====
9
10v9fs is a Unix implementation of the Plan 9 9p remote filesystem protocol.
11
12This software was originally developed by Ron Minnich <rminnich@sandia.gov>
13and Maya Gokhale.  Additional development by Greg Watson
14<gwatson@lanl.gov> and most recently Eric Van Hensbergen
15<ericvh@gmail.com>, Latchesar Ionkov <lucho@ionkov.net> and Russ Cox
16<rsc@swtch.com>.
17
18The best detailed explanation of the Linux implementation and applications of
19the 9p client is available in the form of a USENIX paper:
20
21   https://www.usenix.org/events/usenix05/tech/freenix/hensbergen.html
22
23Other applications are described in the following papers:
24
25	* XCPU & Clustering
26	  http://xcpu.org/papers/xcpu-talk.pdf
27	* KVMFS: control file system for KVM
28	  http://xcpu.org/papers/kvmfs.pdf
29	* CellFS: A New Programming Model for the Cell BE
30	  http://xcpu.org/papers/cellfs-talk.pdf
31	* PROSE I/O: Using 9p to enable Application Partitions
32	  http://plan9.escet.urjc.es/iwp9/cready/PROSE_iwp9_2006.pdf
33	* VirtFS: A Virtualization Aware File System pass-through
34	  https://kernel.org/doc/ols/2010/ols2010-pages-109-120.pdf
35
36Usage
37=====
38
39For remote file server::
40
41	mount -t 9p 10.10.1.2 /mnt/9
42
43For Plan 9 From User Space applications (http://swtch.com/plan9)::
44
45	mount -t 9p `namespace`/acme /mnt/9 -o trans=unix,uname=$USER
46
47For server running on QEMU host with virtio transport::
48
49	mount -t 9p -o trans=virtio <mount_tag> /mnt/9
50
51where mount_tag is the tag generated by the server to each of the exported
52mount points. Each 9P export is seen by the client as a virtio device with an
53associated "mount_tag" property. Available mount tags can be
54seen by reading /sys/bus/virtio/drivers/9pnet_virtio/virtio<n>/mount_tag files.
55
56USBG Usage
57==========
58
59To mount a 9p FS on a USB Host accessible via the gadget at runtime::
60
61	mount -t 9p -o trans=usbg,aname=/path/to/fs <device> /mnt/9
62
63To mount a 9p FS on a USB Host accessible via the gadget as root filesystem::
64
65	root=<device> rootfstype=9p rootflags=trans=usbg,cache=loose,uname=root,access=0,dfltuid=0,dfltgid=0,aname=/path/to/rootfs
66
67where <device> is the tag associated by the usb gadget transport.
68It is defined by the configfs instance name.
69
70USBG Example
71============
72
73The USB host exports a filesystem, while the gadget on the USB device
74side makes it mountable.
75
76Diod (9pfs server) and the forwarder are on the development host, where
77the root filesystem is actually stored. The gadget is initialized during
78boot (or later) on the embedded board. Then the forwarder will find it
79on the USB bus and start forwarding requests.
80
81In this case the 9p requests come from the device and are handled by the
82host. The reason is that USB device ports are normally not available on
83PCs, so a connection in the other direction would not work.
84
85When using the usbg transport, for now there is no native usb host
86service capable to handle the requests from the gadget driver. For
87this we have to use the extra python tool p9_fwd.py from tools/usb.
88
89Just start the 9pfs capable network server like diod/nfs-ganesha e.g.::
90
91        $ diod -f -n -d 0 -S -l 0.0.0.0:9999 -e $PWD
92
93Optionaly scan your bus if there are more then one usbg gadgets to find their path::
94
95        $ python $kernel_dir/tools/usb/p9_fwd.py list
96
97        Bus | Addr | Manufacturer     | Product          | ID        | Path
98        --- | ---- | ---------------- | ---------------- | --------- | ----
99          2 |   67 | unknown          | unknown          | 1d6b:0109 | 2-1.1.2
100          2 |   68 | unknown          | unknown          | 1d6b:0109 | 2-1.1.3
101
102Then start the python transport::
103
104        $ python $kernel_dir/tools/usb/p9_fwd.py --path 2-1.1.2 connect -p 9999
105
106After that the gadget driver can be used as described above.
107
108One use-case is to use it as an alternative to NFS root booting during
109the development of embedded Linux devices.
110
111Options
112=======
113
114  ============= ===============================================================
115  trans=name	select an alternative transport.  Valid options are
116  		currently:
117
118			========  ============================================
119			unix 	  specifying a named pipe mount point
120			tcp	  specifying a normal TCP/IP connection
121			fd   	  used passed file descriptors for connection
122                                  (see rfdno and wfdno)
123			virtio	  connect to the next virtio channel available
124				  (from QEMU with trans_virtio module)
125			rdma	  connect to a specified RDMA channel
126			usbg	  connect to a specified usb gadget channel
127			========  ============================================
128
129  uname=name	user name to attempt mount as on the remote server.  The
130  		server may override or ignore this value.  Certain user
131		names may require authentication.
132
133  aname=name	aname specifies the file tree to access when the server is
134  		offering several exported file systems.
135
136  cache=mode	specifies a caching policy.  By default, no caches are used.
137		The mode can be specified as a bitmask or by using one of the
138		preexisting common 'shortcuts'.
139		The bitmask is described below: (unspecified bits are reserved)
140
141			==========  ====================================================
142			0b00000000  all caches disabled, mmap disabled
143			0b00000001  file caches enabled
144			0b00000010  meta-data caches enabled
145			0b00000100  writeback behavior (as opposed to writethrough)
146			0b00001000  loose caches (no explicit consistency with server)
147			0b10000000  fscache enabled for persistent caching
148			==========  ====================================================
149
150		The current shortcuts and their associated bitmask are:
151
152			=========   ====================================================
153			none        0b00000000 (no caching)
154			readahead   0b00000001 (only read-ahead file caching)
155			mmap        0b00000101 (read-ahead + writeback file cache)
156			loose       0b00001111 (non-coherent file and meta-data caches)
157			fscache     0b10001111 (persistent loose cache)
158			=========   ====================================================
159
160		NOTE: only these shortcuts are tested modes of operation at the
161		moment, so using other combinations of bit-patterns is not
162		known to work.  Work on better cache support is in progress.
163
164		IMPORTANT: loose caches (and by extension at the moment fscache)
165		do not necessarily validate cached values on the server.  In other
166		words changes on the server are not guaranteed to be reflected
167		on the client system.  Only use this mode of operation if you
168		have an exclusive mount and the server will modify the filesystem
169		underneath you.
170
171  debug=n	specifies debug level.  The debug level is a bitmask.
172
173			=====   ================================
174			0x01    display verbose error messages
175			0x02    developer debug (DEBUG_CURRENT)
176			0x04    display 9p trace
177			0x08    display VFS trace
178			0x10    display Marshalling debug
179			0x20    display RPC debug
180			0x40    display transport debug
181			0x80    display allocation debug
182			0x100   display protocol message debug
183			0x200   display Fid debug
184			0x400   display packet debug
185			0x800   display fscache tracing debug
186			=====   ================================
187
188  rfdno=n	the file descriptor for reading with trans=fd
189
190  wfdno=n	the file descriptor for writing with trans=fd
191
192  msize=n	the number of bytes to use for 9p packet payload
193
194  port=n	port to connect to on the remote server
195
196  noextend	force legacy mode (no 9p2000.u or 9p2000.L semantics)
197
198  version=name	Select 9P protocol version. Valid options are:
199
200			========        ==============================
201			9p2000          Legacy mode (same as noextend)
202			9p2000.u        Use 9P2000.u protocol
203			9p2000.L        Use 9P2000.L protocol
204			========        ==============================
205
206  dfltuid	attempt to mount as a particular uid
207
208  dfltgid	attempt to mount with a particular gid
209
210  afid		security channel - used by Plan 9 authentication protocols
211
212  nodevmap	do not map special files - represent them as normal files.
213  		This can be used to share devices/named pipes/sockets between
214		hosts.  This functionality will be expanded in later versions.
215
216  directio	bypass page cache on all read/write operations
217
218  ignoreqv	ignore qid.version==0 as a marker to ignore cache
219
220  noxattr	do not offer xattr functions on this mount.
221
222  access	there are four access modes.
223			user
224				if a user tries to access a file on v9fs
225			        filesystem for the first time, v9fs sends an
226			        attach command (Tattach) for that user.
227				This is the default mode.
228			<uid>
229				allows only user with uid=<uid> to access
230				the files on the mounted filesystem
231			any
232				v9fs does single attach and performs all
233				operations as one user
234			clien
235				 ACL based access check on the 9p client
236			         side for access validation
237
238  cachetag	cache tag to use the specified persistent cache.
239		cache tags for existing cache sessions can be listed at
240		/sys/fs/9p/caches. (applies only to cache=fscache)
241  ============= ===============================================================
242
243Behavior
244========
245
246This section aims at describing 9p 'quirks' that can be different
247from a local filesystem behaviors.
248
249 - Setting O_NONBLOCK on a file will make client reads return as early
250   as the server returns some data instead of trying to fill the read
251   buffer with the requested amount of bytes or end of file is reached.
252
253Resources
254=========
255
256Protocol specifications are maintained on github:
257http://ericvh.github.com/9p-rfc/
258
2599p client and server implementations are listed on
260http://9p.cat-v.org/implementations
261
262A 9p2000.L server is being developed by LLNL and can be found
263at http://code.google.com/p/diod/
264
265There are user and developer mailing lists available through the v9fs project
266on sourceforge (http://sourceforge.net/projects/v9fs).
267
268News and other information is maintained on a Wiki.
269(http://sf.net/apps/mediawiki/v9fs/index.php).
270
271Bug reports are best issued via the mailing list.
272
273For more information on the Plan 9 Operating System check out
274http://plan9.bell-labs.com/plan9
275
276For information on Plan 9 from User Space (Plan 9 applications and libraries
277ported to Linux/BSD/OSX/etc) check out https://9fans.github.io/plan9port/
278