pcmcia: do not use io_req_t after call to pcmcia_request_io()

[linux-2.6-block.git] / Documentation / pcmcia / driver-changes.txt

This file details changes in 2.6 which affect PCMCIA card driver authors:
* No dev_info_t, no cs_types.h (as of 2.6.36)
   dev_info_t and a few other typedefs are removed. No longer use them
   in PCMCIA device drivers. Also, do not include pcmcia/cs_types.h, as
   this file is gone.

* No dev_node_t (as of 2.6.35)
   There is no more need to fill out a "dev_node_t" structure.

* New IRQ request rules (as of 2.6.35)
   Instead of the old pcmcia_request_irq() interface, drivers may now
   choose between:
   - calling request_irq/free_irq directly. Use the IRQ from *p_dev->irq.
   - use pcmcia_request_irq(p_dev, handler_t); the PCMCIA core will
     clean up automatically on calls to pcmcia_disable_device() or
     device ejection.
   - drivers still not capable of IRQF_SHARED (or not telling us so) may
     use the deprecated pcmcia_request_exclusive_irq() for the time
     being; they might receive a shared IRQ nonetheless.

* no cs_error / CS_CHECK / CONFIG_PCMCIA_DEBUG (as of 2.6.33)
   Instead of the cs_error() callback or the CS_CHECK() macro, please use
   Linux-style checking of return values, and -- if necessary -- debug
   messages using "dev_dbg()" or "pr_debug()".

* New CIS tuple access (as of 2.6.33)
   Instead of pcmcia_get_{first,next}_tuple(), pcmcia_get_tuple_data() and
   pcmcia_parse_tuple(), a driver shall use "pcmcia_get_tuple()" if it is
   only interested in one (raw) tuple, or "pcmcia_loop_tuple()" if it is
   interested in all tuples of one type. To decode the MAC from CISTPL_FUNCE,
   a new helper "pcmcia_get_mac_from_cis()" was added.

* New configuration loop helper (as of 2.6.28)
   By calling pcmcia_loop_config(), a driver can iterate over all available
   configuration options. During a driver's probe() phase, one doesn't need
   to use pcmcia_get_{first,next}_tuple, pcmcia_get_tuple_data and
   pcmcia_parse_tuple directly in most if not all cases.

* New release helper (as of 2.6.17)
   Instead of calling pcmcia_release_{configuration,io,irq,win}, all that's
   necessary now is calling pcmcia_disable_device. As there is no valid
   reason left to call pcmcia_release_io and pcmcia_release_irq, the
   exports for them were removed.

* Unify detach and REMOVAL event code, as well as attach and INSERTION
  code (as of 2.6.16)
       void (*remove)          (struct pcmcia_device *dev);
       int (*probe)            (struct pcmcia_device *dev);

* Move suspend, resume and reset out of event handler (as of 2.6.16)
       int (*suspend)          (struct pcmcia_device *dev);
       int (*resume)           (struct pcmcia_device *dev);
  should be initialized in struct pcmcia_driver, and handle
  (SUSPEND == RESET_PHYSICAL) and (RESUME == CARD_RESET) events

* event handler initialization in struct pcmcia_driver (as of 2.6.13)
   The event handler is notified of all events, and must be initialized
   as the event() callback in the driver's struct pcmcia_driver.

* pcmcia/version.h should not be used (as of 2.6.13)
   This file will be removed eventually.

* in-kernel device<->driver matching (as of 2.6.13)
   PCMCIA devices and their correct drivers can now be matched in
   kernelspace. See 'devicetable.txt' for details.

* Device model integration (as of 2.6.11)
   A struct pcmcia_device is registered with the device model core,
   and can be used (e.g. for SET_NETDEV_DEV) by using
   handle_to_dev(client_handle_t * handle).

* Convert internal I/O port addresses to unsigned int (as of 2.6.11)
   ioaddr_t should be replaced by unsigned int in PCMCIA card drivers.

* irq_mask and irq_list parameters (as of 2.6.11)
   The irq_mask and irq_list parameters should no longer be used in
   PCMCIA card drivers. Instead, it is the job of the PCMCIA core to
   determine which IRQ should be used. Therefore, link->irq.IRQInfo2
   is ignored.

* client->PendingEvents is gone (as of 2.6.11)
   client->PendingEvents is no longer available.

* client->Attributes are gone (as of 2.6.11)
   client->Attributes is unused, therefore it is removed from all
   PCMCIA card drivers

* core functions no longer available (as of 2.6.11)
   The following functions have been removed from the kernel source
   because they are unused by all in-kernel drivers, and no external
   driver was reported to rely on them:
	pcmcia_get_first_region()
	pcmcia_get_next_region()
	pcmcia_modify_window()
	pcmcia_set_event_mask()
	pcmcia_get_first_window()
	pcmcia_get_next_window()

* device list iteration upon module removal (as of 2.6.10)
   It is no longer necessary to iterate on the driver's internal
   client list and call the ->detach() function upon module removal.

* Resource management. (as of 2.6.8)
   Although the PCMCIA subsystem will allocate resources for cards,
   it no longer marks these resources busy. This means that driver
   authors are now responsible for claiming your resources as per
   other drivers in Linux. You should use request_region() to mark
   your IO regions in-use, and request_mem_region() to mark your
   memory regions in-use. The name argument should be a pointer to
   your driver name. Eg, for pcnet_cs, name should point to the
   string "pcnet_cs".

* CardServices is gone
  CardServices() in 2.4 is just a big switch statement to call various
  services.  In 2.6, all of those entry points are exported and called
  directly (except for pcmcia_report_error(), just use cs_error() instead).

* struct pcmcia_driver
  You need to use struct pcmcia_driver and pcmcia_{un,}register_driver
  instead of {un,}register_pccard_driver