source: MondoRescue/branches/2.2.9/mindi-busybox/docs/mdev.txt

Last change on this file was 3320, checked in by Bruno Cornec, 9 years ago
  • Re-add (thanks git BTW) the 2.2.9 branch which had been destroyed in the move to 3.0
  • Property svn:eol-style set to native
File size: 5.3 KB
Line 
1-------------
2 MDEV Primer
3-------------
4
5For those of us who know how to use mdev, a primer might seem lame. For
6everyone else, mdev is a weird black box that they hear is awesome, but can't
7seem to get their head around how it works. Thus, a primer.
8
9-----------
10 Basic Use
11-----------
12
13Mdev has two primary uses: initial population and dynamic updates. Both
14require sysfs support in the kernel and have it mounted at /sys. For dynamic
15updates, you also need to have hotplugging enabled in your kernel.
16
17Here's a typical code snippet from the init script:
18[0] mount -t proc proc /proc
19[1] mount -t sysfs sysfs /sys
20[2] echo /sbin/mdev > /proc/sys/kernel/hotplug
21[3] mdev -s
22
23Alternatively, without procfs the above becomes:
24[1] mount -t sysfs sysfs /sys
25[2] sysctl -w kernel.hotplug=/sbin/mdev
26[3] mdev -s
27
28
29Of course, a more "full" setup would entail executing this before the previous
30code snippet:
31[4] mount -t tmpfs -o size=64k,mode=0755 tmpfs /dev
32[5] mkdir /dev/pts
33[6] mount -t devpts devpts /dev/pts
34
35The simple explanation here is that [1] you need to have /sys mounted before
36executing mdev. Then you [2] instruct the kernel to execute /sbin/mdev whenever
37a device is added or removed so that the device node can be created or
38destroyed. Then you [3] seed /dev with all the device nodes that were created
39while the system was booting.
40
41For the "full" setup, you want to [4] make sure /dev is a tmpfs filesystem
42(assuming you're running out of flash). Then you want to [5] create the
43/dev/pts mount point and finally [6] mount the devpts filesystem on it.
44
45-------------
46 MDEV Config (/etc/mdev.conf)
47-------------
48
49Mdev has an optional config file for controlling ownership/permissions of
50device nodes if your system needs something more than the default root/root
51660 permissions.
52
53The file has the format:
54 <device regex> <uid>:<gid> <permissions>
55 or @<maj[,min1[-min2]]> <uid>:<gid> <permissions>
56
57For example:
58 hd[a-z][0-9]* 0:3 660
59
60The config file parsing stops at the first matching line. If no line is
61matched, then the default of 0:0 660 is used. To set your own default, simply
62create your own total match like so:
63 .* 1:1 777
64
65You can rename/move device nodes by using the next optional field.
66 <device regex> <uid>:<gid> <permissions> [=path]
67So if you want to place the device node into a subdirectory, make sure the path
68has a trailing /. If you want to rename the device node, just place the name.
69 hda 0:3 660 =drives/
70This will move "hda" into the drives/ subdirectory.
71 hdb 0:3 660 =cdrom
72This will rename "hdb" to "cdrom".
73
74Similarly, ">path" renames/moves the device but it also creates
75a direct symlink /dev/DEVNAME to the renamed/moved device.
76
77You can also prevent creation of device nodes with the 4th field as "!":
78 tty[a-z]. 0:0 660 !
79 pty[a-z]. 0:0 660 !
80
81If you also enable support for executing your own commands, then the file has
82the format:
83 <device regex> <uid>:<gid> <permissions> [=path] [@|$|*<command>]
84 or
85 <device regex> <uid>:<gid> <permissions> [>path] [@|$|*<command>]
86 or
87 <device regex> <uid>:<gid> <permissions> [!] [@|$|*<command>]
88
89For example:
90---8<---
91# block devices
92([hs]d[a-z]) root:disk 660 >disk/%1/0
93([hs]d[a-z])([0-9]+) root:disk 660 >disk/%1/%2
94mmcblk([0-9]+) root:disk 660 >disk/mmc/%1/0
95mmcblk([0-9]+)p([0-9]+) root:disk 660 >disk/mmc/%1/%2
96# network devices
97(tun|tap) root:network 660 >net/%1
98---8<---
99
100The special characters have the meaning:
101 @ Run after creating the device.
102 $ Run before removing the device.
103 * Run both after creating and before removing the device.
104
105The command is executed via the system() function (which means you're giving a
106command to the shell), so make sure you have a shell installed at /bin/sh. You
107should also keep in mind that the kernel executes hotplug helpers with stdin,
108stdout, and stderr connected to /dev/null.
109
110For your convenience, the shell env var $MDEV is set to the device name. So if
111the device "hdc" was matched, MDEV would be set to "hdc".
112
113----------
114 FIRMWARE
115----------
116
117Some kernel device drivers need to request firmware at runtime in order to
118properly initialize a device. Place all such firmware files into the
119/lib/firmware/ directory. At runtime, the kernel will invoke mdev with the
120filename of the firmware which mdev will load out of /lib/firmware/ and into
121the kernel via the sysfs interface. The exact filename is hardcoded in the
122kernel, so look there if you need to know how to name the file in userspace.
123
124------------
125 SEQUENCING
126------------
127
128Kernel does not serialize hotplug events. It increments SEQNUM environmental
129variable for each successive hotplug invocation. Normally, mdev doesn't care.
130This may reorder hotplug and hot-unplug events, with typical symptoms of
131device nodes sometimes not created as expected.
132
133However, if /dev/mdev.seq file is found, mdev will compare its
134contents with SEQNUM. It will retry up to two seconds, waiting for them
135to match. If they match exactly (not even trailing '\n' is allowed),
136or if two seconds pass, mdev runs as usual, then it rewrites /dev/mdev.seq
137with SEQNUM+1.
138
139IOW: this will serialize concurrent mdev invocations.
140
141If you want to activate this feature, execute "echo >/dev/mdev.seq" prior to
142setting mdev to be the hotplug handler. This writes single '\n' to the file.
143NB: mdev recognizes /dev/mdev.seq consisting of single '\n' character
144as a special case. IOW: this will not make your first hotplug event
145to stall for two seconds.
Note: See TracBrowser for help on using the repository browser.