1 =============================
2 The Linux Kernel Device Model
3 =============================
4
5 Patrick Mochel <mochel@digitalimplant.org>
6
7 Drafted 26 August 2002
8 Updated 31 January 2006
9
10
11 Overview
12 ~~~~~~~~
13
14 The Linux Kernel Driver Model is a unification
15 models that were previously used in the kernel
16 bus-specific drivers for bridges and devices b
17 and operations into globally accessible data s
18
19 Traditional driver models implemented some sor
20 (sometimes just a list) for the devices they c
21 uniformity across the different bus types.
22
23 The current driver model provides a common, un
24 a bus and the devices that can appear under th
25 model includes a set of common attributes whic
26 of common callbacks, such as device discovery
27 shutdown, bus power management, etc.
28
29 The common device and bridge interface reflect
30 computer: namely the ability to do seamless de
31 management, and hot plug. In particular, the m
32 Microsoft (namely ACPI) ensures that almost ev
33 on an x86-compatible system can work within th
34 not every bus is able to support all such oper
35 buses support most of those operations.
36
37
38 Downstream Access
39 ~~~~~~~~~~~~~~~~~
40
41 Common data fields have been moved out of indi
42 data structure. These fields must still be acc
43 and sometimes by the device-specific drivers.
44
45 Other bus layers are encouraged to do what has
46 struct pci_dev now looks like this::
47
48 struct pci_dev {
49 ...
50
51 struct device dev; /* Generic devi
52 ...
53 };
54
55 Note first that the struct device dev within t
56 statically allocated. This means only one allo
57
58 Note also that that struct device dev is not n
59 front of the pci_dev structure. This is to ma
60 they're doing when switching between the bus d
61 and to discourage meaningless and incorrect ca
62
63 The PCI bus layer freely accesses the fields o
64 the structure of struct pci_dev, and it should
65 device. Individual PCI device drivers that hav
66 driver model generally do not and should not t
67 unless there is a compelling reason to do so.
68
69 The above abstraction prevents unnecessary pai
70 If it were not done this way, then when a fiel
71 downstream driver would break. On the other h
72 (and not the device layer) accesses the struct
73 layer that needs to change.
74
75
76 User Interface
77 ~~~~~~~~~~~~~~
78
79 By virtue of having a complete hierarchical vi
80 system, exporting a complete hierarchical view
81 easy. This has been accomplished by implementi
82 file system named sysfs.
83
84 Almost all mainstream Linux distros mount this
85 can see some variation of the following in the
86
87 $ mount
88 ...
89 none on /sys type sysfs (rw,noexec,nosuid,no
90 ...
91 $
92
93 The auto-mounting of sysfs is typically accomp
94 the following in the /etc/fstab file::
95
96 none /sys sysfs defaults
97
98 or something similar in the /lib/init/fstab fi
99
100 none /sys sysfs nodev,noexe
101
102 If sysfs is not automatically mounted, you can
103
104 # mount -t sysfs sysfs /sys
105
106 Whenever a device is inserted into the tree, a
107 This directory may be populated at each layer
108 the bus layer, or the device layer.
109
110 The global layer currently creates two files -
111 former only reports the name of the device. Th
112 current power state of the device. It will als
113 power state.
114
115 The bus layer may also create files for the de
116 bus. For example, the PCI layer currently crea
117 for each PCI device.
118
119 A device-specific driver may also export files
120 device-specific data or tunable interfaces.
121
122 More information about the sysfs directory lay
123 the other documents in this directory and in t
124 Documentation/filesystems/sysfs.rst.
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.