xref: /linux/Documentation/pcmcia/driver-changes.rst (revision 0898782247ae533d1f4e47a06bc5d4870931b284)
1==============
2Driver changes
3==============
4
5This file details changes in 2.6 which affect PCMCIA card driver authors:
6
7* pcmcia_loop_config() and autoconfiguration (as of 2.6.36)
8   If `struct pcmcia_device *p_dev->config_flags` is set accordingly,
9   pcmcia_loop_config() now sets up certain configuration values
10   automatically, though the driver may still override the settings
11   in the callback function. The following autoconfiguration options
12   are provided at the moment:
13
14	- CONF_AUTO_CHECK_VCC : check for matching Vcc
15	- CONF_AUTO_SET_VPP   : set Vpp
16	- CONF_AUTO_AUDIO     : auto-enable audio line, if required
17	- CONF_AUTO_SET_IO    : set ioport resources (->resource[0,1])
18	- CONF_AUTO_SET_IOMEM : set first iomem resource (->resource[2])
19
20* pcmcia_request_configuration -> pcmcia_enable_device (as of 2.6.36)
21   pcmcia_request_configuration() got renamed to pcmcia_enable_device(),
22   as it mirrors pcmcia_disable_device(). Configuration settings are now
23   stored in struct pcmcia_device, e.g. in the fields config_flags,
24   config_index, config_base, vpp.
25
26* pcmcia_request_window changes (as of 2.6.36)
27   Instead of win_req_t, drivers are now requested to fill out
28   `struct pcmcia_device *p_dev->resource[2,3,4,5]` for up to four ioport
29   ranges. After a call to pcmcia_request_window(), the regions found there
30   are reserved and may be used immediately -- until pcmcia_release_window()
31   is called.
32
33* pcmcia_request_io changes (as of 2.6.36)
34   Instead of io_req_t, drivers are now requested to fill out
35   `struct pcmcia_device *p_dev->resource[0,1]` for up to two ioport
36   ranges. After a call to pcmcia_request_io(), the ports found there
37   are reserved, after calling pcmcia_request_configuration(), they may
38   be used.
39
40* No dev_info_t, no cs_types.h (as of 2.6.36)
41   dev_info_t and a few other typedefs are removed. No longer use them
42   in PCMCIA device drivers. Also, do not include pcmcia/cs_types.h, as
43   this file is gone.
44
45* No dev_node_t (as of 2.6.35)
46   There is no more need to fill out a "dev_node_t" structure.
47
48* New IRQ request rules (as of 2.6.35)
49   Instead of the old pcmcia_request_irq() interface, drivers may now
50   choose between:
51
52   - calling request_irq/free_irq directly. Use the IRQ from `*p_dev->irq`.
53   - use pcmcia_request_irq(p_dev, handler_t); the PCMCIA core will
54     clean up automatically on calls to pcmcia_disable_device() or
55     device ejection.
56
57* no cs_error / CS_CHECK / CONFIG_PCMCIA_DEBUG (as of 2.6.33)
58   Instead of the cs_error() callback or the CS_CHECK() macro, please use
59   Linux-style checking of return values, and -- if necessary -- debug
60   messages using "dev_dbg()" or "pr_debug()".
61
62* New CIS tuple access (as of 2.6.33)
63   Instead of pcmcia_get_{first,next}_tuple(), pcmcia_get_tuple_data() and
64   pcmcia_parse_tuple(), a driver shall use "pcmcia_get_tuple()" if it is
65   only interested in one (raw) tuple, or "pcmcia_loop_tuple()" if it is
66   interested in all tuples of one type. To decode the MAC from CISTPL_FUNCE,
67   a new helper "pcmcia_get_mac_from_cis()" was added.
68
69* New configuration loop helper (as of 2.6.28)
70   By calling pcmcia_loop_config(), a driver can iterate over all available
71   configuration options. During a driver's probe() phase, one doesn't need
72   to use pcmcia_get_{first,next}_tuple, pcmcia_get_tuple_data and
73   pcmcia_parse_tuple directly in most if not all cases.
74
75* New release helper (as of 2.6.17)
76   Instead of calling pcmcia_release_{configuration,io,irq,win}, all that's
77   necessary now is calling pcmcia_disable_device. As there is no valid
78   reason left to call pcmcia_release_io and pcmcia_release_irq, the
79   exports for them were removed.
80
81* Unify detach and REMOVAL event code, as well as attach and INSERTION
82  code (as of 2.6.16)::
83
84       void (*remove)          (struct pcmcia_device *dev);
85       int (*probe)            (struct pcmcia_device *dev);
86
87* Move suspend, resume and reset out of event handler (as of 2.6.16)::
88
89       int (*suspend)          (struct pcmcia_device *dev);
90       int (*resume)           (struct pcmcia_device *dev);
91
92  should be initialized in struct pcmcia_driver, and handle
93  (SUSPEND == RESET_PHYSICAL) and (RESUME == CARD_RESET) events
94
95* event handler initialization in struct pcmcia_driver (as of 2.6.13)
96   The event handler is notified of all events, and must be initialized
97   as the event() callback in the driver's struct pcmcia_driver.
98
99* pcmcia/version.h should not be used (as of 2.6.13)
100   This file will be removed eventually.
101
102* in-kernel device<->driver matching (as of 2.6.13)
103   PCMCIA devices and their correct drivers can now be matched in
104   kernelspace. See 'devicetable.txt' for details.
105
106* Device model integration (as of 2.6.11)
107   A struct pcmcia_device is registered with the device model core,
108   and can be used (e.g. for SET_NETDEV_DEV) by using
109   handle_to_dev(client_handle_t * handle).
110
111* Convert internal I/O port addresses to unsigned int (as of 2.6.11)
112   ioaddr_t should be replaced by unsigned int in PCMCIA card drivers.
113
114* irq_mask and irq_list parameters (as of 2.6.11)
115   The irq_mask and irq_list parameters should no longer be used in
116   PCMCIA card drivers. Instead, it is the job of the PCMCIA core to
117   determine which IRQ should be used. Therefore, link->irq.IRQInfo2
118   is ignored.
119
120* client->PendingEvents is gone (as of 2.6.11)
121   client->PendingEvents is no longer available.
122
123* client->Attributes are gone (as of 2.6.11)
124   client->Attributes is unused, therefore it is removed from all
125   PCMCIA card drivers
126
127* core functions no longer available (as of 2.6.11)
128   The following functions have been removed from the kernel source
129   because they are unused by all in-kernel drivers, and no external
130   driver was reported to rely on them::
131
132	pcmcia_get_first_region()
133	pcmcia_get_next_region()
134	pcmcia_modify_window()
135	pcmcia_set_event_mask()
136	pcmcia_get_first_window()
137	pcmcia_get_next_window()
138
139* device list iteration upon module removal (as of 2.6.10)
140   It is no longer necessary to iterate on the driver's internal
141   client list and call the ->detach() function upon module removal.
142
143* Resource management. (as of 2.6.8)
144   Although the PCMCIA subsystem will allocate resources for cards,
145   it no longer marks these resources busy. This means that driver
146   authors are now responsible for claiming your resources as per
147   other drivers in Linux. You should use request_region() to mark
148   your IO regions in-use, and request_mem_region() to mark your
149   memory regions in-use. The name argument should be a pointer to
150   your driver name. Eg, for pcnet_cs, name should point to the
151   string "pcnet_cs".
152
153* CardServices is gone
154  CardServices() in 2.4 is just a big switch statement to call various
155  services.  In 2.6, all of those entry points are exported and called
156  directly (except for pcmcia_report_error(), just use cs_error() instead).
157
158* struct pcmcia_driver
159  You need to use struct pcmcia_driver and pcmcia_{un,}register_driver
160  instead of {un,}register_pccard_driver
161