Based on kernel version 4.16.1. Page generated on 2018-04-09 11:53 EST.
1 The Linux Kernel Device Model 2 3 Patrick Mochel <mochel@digitalimplant.org> 4 5 Drafted 26 August 2002 6 Updated 31 January 2006 7 8 9 Overview 10 ~~~~~~~~ 11 12 The Linux Kernel Driver Model is a unification of all the disparate driver 13 models that were previously used in the kernel. It is intended to augment the 14 bus-specific drivers for bridges and devices by consolidating a set of data 15 and operations into globally accessible data structures. 16 17 Traditional driver models implemented some sort of tree-like structure 18 (sometimes just a list) for the devices they control. There wasn't any 19 uniformity across the different bus types. 20 21 The current driver model provides a common, uniform data model for describing 22 a bus and the devices that can appear under the bus. The unified bus 23 model includes a set of common attributes which all busses carry, and a set 24 of common callbacks, such as device discovery during bus probing, bus 25 shutdown, bus power management, etc. 26 27 The common device and bridge interface reflects the goals of the modern 28 computer: namely the ability to do seamless device "plug and play", power 29 management, and hot plug. In particular, the model dictated by Intel and 30 Microsoft (namely ACPI) ensures that almost every device on almost any bus 31 on an x86-compatible system can work within this paradigm. Of course, 32 not every bus is able to support all such operations, although most 33 buses support most of those operations. 34 35 36 Downstream Access 37 ~~~~~~~~~~~~~~~~~ 38 39 Common data fields have been moved out of individual bus layers into a common 40 data structure. These fields must still be accessed by the bus layers, 41 and sometimes by the device-specific drivers. 42 43 Other bus layers are encouraged to do what has been done for the PCI layer. 44 struct pci_dev now looks like this: 45 46 struct pci_dev { 47 ... 48 49 struct device dev; /* Generic device interface */ 50 ... 51 }; 52 53 Note first that the struct device dev within the struct pci_dev is 54 statically allocated. This means only one allocation on device discovery. 55 56 Note also that that struct device dev is not necessarily defined at the 57 front of the pci_dev structure. This is to make people think about what 58 they're doing when switching between the bus driver and the global driver, 59 and to discourage meaningless and incorrect casts between the two. 60 61 The PCI bus layer freely accesses the fields of struct device. It knows about 62 the structure of struct pci_dev, and it should know the structure of struct 63 device. Individual PCI device drivers that have been converted to the current 64 driver model generally do not and should not touch the fields of struct device, 65 unless there is a compelling reason to do so. 66 67 The above abstraction prevents unnecessary pain during transitional phases. 68 If it were not done this way, then when a field was renamed or removed, every 69 downstream driver would break. On the other hand, if only the bus layer 70 (and not the device layer) accesses the struct device, it is only the bus 71 layer that needs to change. 72 73 74 User Interface 75 ~~~~~~~~~~~~~~ 76 77 By virtue of having a complete hierarchical view of all the devices in the 78 system, exporting a complete hierarchical view to userspace becomes relatively 79 easy. This has been accomplished by implementing a special purpose virtual 80 file system named sysfs. 81 82 Almost all mainstream Linux distros mount this filesystem automatically; you 83 can see some variation of the following in the output of the "mount" command: 84 85 $ mount 86 ... 87 none on /sys type sysfs (rw,noexec,nosuid,nodev) 88 ... 89 $ 90 91 The auto-mounting of sysfs is typically accomplished by an entry similar to 92 the following in the /etc/fstab file: 93 94 none /sys sysfs defaults 0 0 95 96 or something similar in the /lib/init/fstab file on Debian-based systems: 97 98 none /sys sysfs nodev,noexec,nosuid 0 0 99 100 If sysfs is not automatically mounted, you can always do it manually with: 101 102 # mount -t sysfs sysfs /sys 103 104 Whenever a device is inserted into the tree, a directory is created for it. 105 This directory may be populated at each layer of discovery - the global layer, 106 the bus layer, or the device layer. 107 108 The global layer currently creates two files - 'name' and 'power'. The 109 former only reports the name of the device. The latter reports the 110 current power state of the device. It will also be used to set the current 111 power state. 112 113 The bus layer may also create files for the devices it finds while probing the 114 bus. For example, the PCI layer currently creates 'irq' and 'resource' files 115 for each PCI device. 116 117 A device-specific driver may also export files in its directory to expose 118 device-specific data or tunable interfaces. 119 120 More information about the sysfs directory layout can be found in 121 the other documents in this directory and in the file 122 Documentation/filesystems/sysfs.txt.