1Tmpfs is a file system which keeps all files in virtual memory.
2
3
4Everything in tmpfs is temporary in the sense that no files will be
5created on your hard drive. If you unmount a tmpfs instance,
6everything stored therein is lost.
7
8tmpfs puts everything into the kernel internal caches and grows and
9shrinks to accommodate the files it contains and is able to swap
10unneeded pages out to swap space. It has maximum size limits which can
11be adjusted on the fly via 'mount -o remount ...'
12
13If you compare it to ramfs (which was the template to create tmpfs)
14you gain swapping and limit checking. Another similar thing is the RAM
15disk (/dev/ram*), which simulates a fixed size hard disk in physical
16RAM, where you have to create an ordinary filesystem on top. Ramdisks
17cannot swap and you do not have the possibility to resize them. 
18
19Since tmpfs lives completely in the page cache and on swap, all tmpfs
20pages currently in memory will show up as cached. It will not show up
21as shared or something like that. Further on you can check the actual
22RAM+swap use of a tmpfs instance with df(1) and du(1).
23
24
25tmpfs has the following uses:
26
271) There is always a kernel internal mount which you will not see at
28   all. This is used for shared anonymous mappings and SYSV shared
29   memory. 
30
31   This mount does not depend on CONFIG_TMPFS. If CONFIG_TMPFS is not
32   set, the user visible part of tmpfs is not build. But the internal
33   mechanisms are always present.
34
352) glibc 2.2 and above expects tmpfs to be mounted at /dev/shm for
36   POSIX shared memory (shm_open, shm_unlink). Adding the following
37   line to /etc/fstab should take care of this:
38
39	tmpfs	/dev/shm	tmpfs	defaults	0 0
40
41   Remember to create the directory that you intend to mount tmpfs on
42   if necessary.
43
44   This mount is _not_ needed for SYSV shared memory. The internal
45   mount is used for that. (In the 2.3 kernel versions it was
46   necessary to mount the predecessor of tmpfs (shm fs) to use SYSV
47   shared memory)
48
493) Some people (including me) find it very convenient to mount it
50   e.g. on /tmp and /var/tmp and have a big swap partition. And now
51   loop mounts of tmpfs files do work, so mkinitrd shipped by most
52   distributions should succeed with a tmpfs /tmp.
53
544) And probably a lot more I do not know about :-)
55
56
57tmpfs has three mount options for sizing:
58
59size:      The limit of allocated bytes for this tmpfs instance. The 
60           default is half of your physical RAM without swap. If you
61           oversize your tmpfs instances the machine will deadlock
62           since the OOM handler will not be able to free that memory.
63nr_blocks: The same as size, but in blocks of PAGE_CACHE_SIZE.
64nr_inodes: The maximum number of inodes for this instance. The default
65           is half of the number of your physical RAM pages, or (on a
66           machine with highmem) the number of lowmem RAM pages,
67           whichever is the lower.
68
69These parameters accept a suffix k, m or g for kilo, mega and giga and
70can be changed on remount.  The size parameter also accepts a suffix %
71to limit this tmpfs instance to that percentage of your physical RAM:
72the default, when neither size nor nr_blocks is specified, is size=50%
73
74If nr_blocks=0 (or size=0), blocks will not be limited in that instance;
75if nr_inodes=0, inodes will not be limited.  It is generally unwise to
76mount with such options, since it allows any user with write access to
77use up all the memory on the machine; but enhances the scalability of
78that instance in a system with many cpus making intensive use of it.
79
80
81tmpfs has a mount option to set the NUMA memory allocation policy for
82all files in that instance (if CONFIG_NUMA is enabled) - which can be
83adjusted on the fly via 'mount -o remount ...'
84
85mpol=default             prefers to allocate memory from the local node
86mpol=prefer:Node         prefers to allocate memory from the given Node
87mpol=bind:NodeList       allocates memory only from nodes in NodeList
88mpol=interleave          prefers to allocate from each node in turn
89mpol=interleave:NodeList allocates from each node of NodeList in turn
90
91NodeList format is a comma-separated list of decimal numbers and ranges,
92a range being two hyphen-separated decimal numbers, the smallest and
93largest node numbers in the range.  For example, mpol=bind:0-3,5,7,9-15
94
95Note that trying to mount a tmpfs with an mpol option will fail if the
96running kernel does not support NUMA; and will fail if its nodelist
97specifies a node which is not online.  If your system relies on that
98tmpfs being mounted, but from time to time runs a kernel built without
99NUMA capability (perhaps a safe recovery kernel), or with fewer nodes
100online, then it is advisable to omit the mpol option from automatic
101mount options.  It can be added later, when the tmpfs is already mounted
102on MountPoint, by 'mount -o remount,mpol=Policy:NodeList MountPoint'.
103
104
105To specify the initial root directory you can use the following mount
106options:
107
108mode:	The permissions as an octal number
109uid:	The user id 
110gid:	The group id
111
112These options do not have any effect on remount. You can change these
113parameters with chmod(1), chown(1) and chgrp(1) on a mounted filesystem.
114
115
116So 'mount -t tmpfs -o size=10G,nr_inodes=10k,mode=700 tmpfs /mytmpfs'
117will give you tmpfs instance on /mytmpfs which can allocate 10GB
118RAM/SWAP in 10240 inodes and it is only accessible by root.
119
120
121Author:
122   Christoph Rohland <cr@sap.com>, 1.12.01
123Updated:
124   Hugh Dickins <hugh@veritas.com>, 4 June 2007
125