Commit | Line | Data |
---|---|---|
5085cb26 | 1 | This file details changes in 2.6 which affect PCMCIA card driver authors: |
90abdc3b DB |
2 | * pcmcia_request_io changes (as of 2.6.36) |
3 | Instead of io_req_t, drivers are now requested to fill out | |
4 | struct pcmcia_device *p_dev->resource[0,1] for up to two ioport | |
5 | ranges. After a call to pcmcia_request_io(), the ports found there | |
6 | are reserved, after calling pcmcia_request_configuration(), they may | |
7 | be used. | |
8 | ||
ac8b4228 DB |
9 | * No dev_info_t, no cs_types.h (as of 2.6.36) |
10 | dev_info_t and a few other typedefs are removed. No longer use them | |
11 | in PCMCIA device drivers. Also, do not include pcmcia/cs_types.h, as | |
12 | this file is gone. | |
13 | ||
b9300aa7 DB |
14 | * No dev_node_t (as of 2.6.35) |
15 | There is no more need to fill out a "dev_node_t" structure. | |
16 | ||
eb14120f DB |
17 | * New IRQ request rules (as of 2.6.35) |
18 | Instead of the old pcmcia_request_irq() interface, drivers may now | |
19 | choose between: | |
20 | - calling request_irq/free_irq directly. Use the IRQ from *p_dev->irq. | |
21 | - use pcmcia_request_irq(p_dev, handler_t); the PCMCIA core will | |
22 | clean up automatically on calls to pcmcia_disable_device() or | |
23 | device ejection. | |
24 | - drivers still not capable of IRQF_SHARED (or not telling us so) may | |
25 | use the deprecated pcmcia_request_exclusive_irq() for the time | |
26 | being; they might receive a shared IRQ nonetheless. | |
5085cb26 | 27 | |
9cb495bb DB |
28 | * no cs_error / CS_CHECK / CONFIG_PCMCIA_DEBUG (as of 2.6.33) |
29 | Instead of the cs_error() callback or the CS_CHECK() macro, please use | |
30 | Linux-style checking of return values, and -- if necessary -- debug | |
31 | messages using "dev_dbg()" or "pr_debug()". | |
32 | ||
91284224 DB |
33 | * New CIS tuple access (as of 2.6.33) |
34 | Instead of pcmcia_get_{first,next}_tuple(), pcmcia_get_tuple_data() and | |
35 | pcmcia_parse_tuple(), a driver shall use "pcmcia_get_tuple()" if it is | |
36 | only interested in one (raw) tuple, or "pcmcia_loop_tuple()" if it is | |
37 | interested in all tuples of one type. To decode the MAC from CISTPL_FUNCE, | |
38 | a new helper "pcmcia_get_mac_from_cis()" was added. | |
39 | ||
a804b574 DB |
40 | * New configuration loop helper (as of 2.6.28) |
41 | By calling pcmcia_loop_config(), a driver can iterate over all available | |
42 | configuration options. During a driver's probe() phase, one doesn't need | |
43 | to use pcmcia_get_{first,next}_tuple, pcmcia_get_tuple_data and | |
44 | pcmcia_parse_tuple directly in most if not all cases. | |
45 | ||
5f2a71fc DB |
46 | * New release helper (as of 2.6.17) |
47 | Instead of calling pcmcia_release_{configuration,io,irq,win}, all that's | |
48 | necessary now is calling pcmcia_disable_device. As there is no valid | |
50db3fdb DB |
49 | reason left to call pcmcia_release_io and pcmcia_release_irq, the |
50 | exports for them were removed. | |
5f2a71fc | 51 | |
f8cfa618 DB |
52 | * Unify detach and REMOVAL event code, as well as attach and INSERTION |
53 | code (as of 2.6.16) | |
54 | void (*remove) (struct pcmcia_device *dev); | |
55 | int (*probe) (struct pcmcia_device *dev); | |
cc3b4866 | 56 | |
98e4c28b DB |
57 | * Move suspend, resume and reset out of event handler (as of 2.6.16) |
58 | int (*suspend) (struct pcmcia_device *dev); | |
59 | int (*resume) (struct pcmcia_device *dev); | |
60 | should be initialized in struct pcmcia_driver, and handle | |
61 | (SUSPEND == RESET_PHYSICAL) and (RESUME == CARD_RESET) events | |
62 | ||
c6fd7188 DB |
63 | * event handler initialization in struct pcmcia_driver (as of 2.6.13) |
64 | The event handler is notified of all events, and must be initialized | |
65 | as the event() callback in the driver's struct pcmcia_driver. | |
66 | ||
67 | * pcmcia/version.h should not be used (as of 2.6.13) | |
68 | This file will be removed eventually. | |
69 | ||
70 | * in-kernel device<->driver matching (as of 2.6.13) | |
5085cb26 DB |
71 | PCMCIA devices and their correct drivers can now be matched in |
72 | kernelspace. See 'devicetable.txt' for details. | |
73 | ||
74 | * Device model integration (as of 2.6.11) | |
75 | A struct pcmcia_device is registered with the device model core, | |
76 | and can be used (e.g. for SET_NETDEV_DEV) by using | |
77 | handle_to_dev(client_handle_t * handle). | |
78 | ||
906da809 OJ |
79 | * Convert internal I/O port addresses to unsigned int (as of 2.6.11) |
80 | ioaddr_t should be replaced by unsigned int in PCMCIA card drivers. | |
5085cb26 DB |
81 | |
82 | * irq_mask and irq_list parameters (as of 2.6.11) | |
83 | The irq_mask and irq_list parameters should no longer be used in | |
84 | PCMCIA card drivers. Instead, it is the job of the PCMCIA core to | |
85 | determine which IRQ should be used. Therefore, link->irq.IRQInfo2 | |
86 | is ignored. | |
87 | ||
88 | * client->PendingEvents is gone (as of 2.6.11) | |
89 | client->PendingEvents is no longer available. | |
90 | ||
91 | * client->Attributes are gone (as of 2.6.11) | |
92 | client->Attributes is unused, therefore it is removed from all | |
93 | PCMCIA card drivers | |
94 | ||
95 | * core functions no longer available (as of 2.6.11) | |
96 | The following functions have been removed from the kernel source | |
97 | because they are unused by all in-kernel drivers, and no external | |
98 | driver was reported to rely on them: | |
99 | pcmcia_get_first_region() | |
100 | pcmcia_get_next_region() | |
101 | pcmcia_modify_window() | |
102 | pcmcia_set_event_mask() | |
103 | pcmcia_get_first_window() | |
104 | pcmcia_get_next_window() | |
105 | ||
106 | * device list iteration upon module removal (as of 2.6.10) | |
107 | It is no longer necessary to iterate on the driver's internal | |
108 | client list and call the ->detach() function upon module removal. | |
109 | ||
110 | * Resource management. (as of 2.6.8) | |
111 | Although the PCMCIA subsystem will allocate resources for cards, | |
112 | it no longer marks these resources busy. This means that driver | |
113 | authors are now responsible for claiming your resources as per | |
114 | other drivers in Linux. You should use request_region() to mark | |
115 | your IO regions in-use, and request_mem_region() to mark your | |
116 | memory regions in-use. The name argument should be a pointer to | |
117 | your driver name. Eg, for pcnet_cs, name should point to the | |
118 | string "pcnet_cs". | |
661d04c6 DB |
119 | |
120 | * CardServices is gone | |
121 | CardServices() in 2.4 is just a big switch statement to call various | |
122 | services. In 2.6, all of those entry points are exported and called | |
123 | directly (except for pcmcia_report_error(), just use cs_error() instead). | |
124 | ||
125 | * struct pcmcia_driver | |
126 | You need to use struct pcmcia_driver and pcmcia_{un,}register_driver | |
127 | instead of {un,}register_pccard_driver |