atyfb: Properly save PCI state before changing PCI PM level
[deliverable/linux.git] / Documentation / pnp.txt
1 Linux Plug and Play Documentation
2 by Adam Belay <ambx1@neo.rr.com>
3 last updated: Oct. 16, 2002
4 ---------------------------------------------------------------------------------------
5
6
7
8 Overview
9 --------
10 Plug and Play provides a means of detecting and setting resources for legacy or
11 otherwise unconfigurable devices. The Linux Plug and Play Layer provides these
12 services to compatible drivers.
13
14
15
16 The User Interface
17 ------------------
18 The Linux Plug and Play user interface provides a means to activate PnP devices
19 for legacy and user level drivers that do not support Linux Plug and Play. The
20 user interface is integrated into sysfs.
21
22 In addition to the standard sysfs file the following are created in each
23 device's directory:
24 id - displays a list of support EISA IDs
25 options - displays possible resource configurations
26 resources - displays currently allocated resources and allows resource changes
27
28 -activating a device
29
30 #echo "auto" > resources
31
32 this will invoke the automatic resource config system to activate the device
33
34 -manually activating a device
35
36 #echo "manual <depnum> <mode>" > resources
37 <depnum> - the configuration number
38 <mode> - static or dynamic
39 static = for next boot
40 dynamic = now
41
42 -disabling a device
43
44 #echo "disable" > resources
45
46
47 EXAMPLE:
48
49 Suppose you need to activate the floppy disk controller.
50 1.) change to the proper directory, in my case it is
51 /driver/bus/pnp/devices/00:0f
52 # cd /driver/bus/pnp/devices/00:0f
53 # cat name
54 PC standard floppy disk controller
55
56 2.) check if the device is already active
57 # cat resources
58 DISABLED
59
60 - Notice the string "DISABLED". THis means the device is not active.
61
62 3.) check the device's possible configurations (optional)
63 # cat options
64 Dependent: 01 - Priority acceptable
65 port 0x3f0-0x3f0, align 0x7, size 0x6, 16-bit address decoding
66 port 0x3f7-0x3f7, align 0x0, size 0x1, 16-bit address decoding
67 irq 6
68 dma 2 8-bit compatible
69 Dependent: 02 - Priority acceptable
70 port 0x370-0x370, align 0x7, size 0x6, 16-bit address decoding
71 port 0x377-0x377, align 0x0, size 0x1, 16-bit address decoding
72 irq 6
73 dma 2 8-bit compatible
74
75 4.) now activate the device
76 # echo "auto" > resources
77
78 5.) finally check if the device is active
79 # cat resources
80 io 0x3f0-0x3f5
81 io 0x3f7-0x3f7
82 irq 6
83 dma 2
84
85 also there are a series of kernel parameters:
86 pnp_reserve_irq=irq1[,irq2] ....
87 pnp_reserve_dma=dma1[,dma2] ....
88 pnp_reserve_io=io1,size1[,io2,size2] ....
89 pnp_reserve_mem=mem1,size1[,mem2,size2] ....
90
91
92
93 The Unified Plug and Play Layer
94 -------------------------------
95 All Plug and Play drivers, protocols, and services meet at a central location
96 called the Plug and Play Layer. This layer is responsible for the exchange of
97 information between PnP drivers and PnP protocols. Thus it automatically
98 forwards commands to the proper protocol. This makes writing PnP drivers
99 significantly easier.
100
101 The following functions are available from the Plug and Play Layer:
102
103 pnp_get_protocol
104 - increments the number of uses by one
105
106 pnp_put_protocol
107 - deincrements the number of uses by one
108
109 pnp_register_protocol
110 - use this to register a new PnP protocol
111
112 pnp_unregister_protocol
113 - use this function to remove a PnP protocol from the Plug and Play Layer
114
115 pnp_register_driver
116 - adds a PnP driver to the Plug and Play Layer
117 - this includes driver model integration
118 - returns zero for success or a negative error number for failure; count
119 calls to the .add() method if you need to know how many devices bind to
120 the driver
121
122 pnp_unregister_driver
123 - removes a PnP driver from the Plug and Play Layer
124
125
126
127 Plug and Play Protocols
128 -----------------------
129 This section contains information for PnP protocol developers.
130
131 The following Protocols are currently available in the computing world:
132 - PNPBIOS: used for system devices such as serial and parallel ports.
133 - ISAPNP: provides PnP support for the ISA bus
134 - ACPI: among its many uses, ACPI provides information about system level
135 devices.
136 It is meant to replace the PNPBIOS. It is not currently supported by Linux
137 Plug and Play but it is planned to be in the near future.
138
139
140 Requirements for a Linux PnP protocol:
141 1.) the protocol must use EISA IDs
142 2.) the protocol must inform the PnP Layer of a devices current configuration
143 - the ability to set resources is optional but preferred.
144
145 The following are PnP protocol related functions:
146
147 pnp_add_device
148 - use this function to add a PnP device to the PnP layer
149 - only call this function when all wanted values are set in the pnp_dev
150 structure
151
152 pnp_init_device
153 - call this to initialize the PnP structure
154
155 pnp_remove_device
156 - call this to remove a device from the Plug and Play Layer.
157 - it will fail if the device is still in use.
158 - automatically will free mem used by the device and related structures
159
160 pnp_add_id
161 - adds a EISA ID to the list of supported IDs for the specified device
162
163 For more information consult the source of a protocol such as
164 /drivers/pnp/pnpbios/core.c.
165
166
167
168 Linux Plug and Play Drivers
169 ---------------------------
170 This section contains information for linux PnP driver developers.
171
172 The New Way
173 ...........
174 1.) first make a list of supported EISA IDS
175 ex:
176 static const struct pnp_id pnp_dev_table[] = {
177 /* Standard LPT Printer Port */
178 {.id = "PNP0400", .driver_data = 0},
179 /* ECP Printer Port */
180 {.id = "PNP0401", .driver_data = 0},
181 {.id = ""}
182 };
183
184 Please note that the character 'X' can be used as a wild card in the function
185 portion (last four characters).
186 ex:
187 /* Unknown PnP modems */
188 { "PNPCXXX", UNKNOWN_DEV },
189
190 Supported PnP card IDs can optionally be defined.
191 ex:
192 static const struct pnp_id pnp_card_table[] = {
193 { "ANYDEVS", 0 },
194 { "", 0 }
195 };
196
197 2.) Optionally define probe and remove functions. It may make sense not to
198 define these functions if the driver already has a reliable method of detecting
199 the resources, such as the parport_pc driver.
200 ex:
201 static int
202 serial_pnp_probe(struct pnp_dev * dev, const struct pnp_id *card_id, const
203 struct pnp_id *dev_id)
204 {
205 . . .
206
207 ex:
208 static void serial_pnp_remove(struct pnp_dev * dev)
209 {
210 . . .
211
212 consult /drivers/serial/8250_pnp.c for more information.
213
214 3.) create a driver structure
215 ex:
216
217 static struct pnp_driver serial_pnp_driver = {
218 .name = "serial",
219 .card_id_table = pnp_card_table,
220 .id_table = pnp_dev_table,
221 .probe = serial_pnp_probe,
222 .remove = serial_pnp_remove,
223 };
224
225 * name and id_table cannot be NULL.
226
227 4.) register the driver
228 ex:
229
230 static int __init serial8250_pnp_init(void)
231 {
232 return pnp_register_driver(&serial_pnp_driver);
233 }
234
235 The Old Way
236 ...........
237
238 a series of compatibility functions have been created to make it easy to convert
239
240 ISAPNP drivers. They should serve as a temporary solution only.
241
242 they are as follows:
243
244 struct pnp_card *pnp_find_card(unsigned short vendor,
245 unsigned short device,
246 struct pnp_card *from)
247
248 struct pnp_dev *pnp_find_dev(struct pnp_card *card,
249 unsigned short vendor,
250 unsigned short function,
251 struct pnp_dev *from)
252
This page took 0.037222 seconds and 5 git commands to generate.