1This file details changes in 2.6 which affect PCMCIA card driver authors:
   2* pcmcia_loop_config() and autoconfiguration (as of 2.6.36)
   3   If struct pcmcia_device *p_dev->config_flags is set accordingly,
   4   pcmcia_loop_config() now sets up certain configuration values
   5   automatically, though the driver may still override the settings
   6   in the callback function. The following autoconfiguration options
   7   are provided at the moment:
   8        CONF_AUTO_CHECK_VCC : check for matching Vcc
   9        CONF_AUTO_SET_VPP   : set Vpp
  10        CONF_AUTO_AUDIO     : auto-enable audio line, if required
  11        CONF_AUTO_SET_IO    : set ioport resources (->resource[0,1])
  12        CONF_AUTO_SET_IOMEM : set first iomem resource (->resource[2])
  14* pcmcia_request_configuration -> pcmcia_enable_device (as of 2.6.36)
  15   pcmcia_request_configuration() got renamed to pcmcia_enable_device(),
  16   as it mirrors pcmcia_disable_device(). Configuration settings are now
  17   stored in struct pcmcia_device, e.g. in the fields config_flags,
  18   config_index, config_base, vpp.
  20* pcmcia_request_window changes (as of 2.6.36)
  21   Instead of win_req_t, drivers are now requested to fill out
  22   struct pcmcia_device *p_dev->resource[2,3,4,5] for up to four ioport
  23   ranges. After a call to pcmcia_request_window(), the regions found there
  24   are reserved and may be used immediately -- until pcmcia_release_window()
  25   is called.
  27* pcmcia_request_io changes (as of 2.6.36)
  28   Instead of io_req_t, drivers are now requested to fill out
  29   struct pcmcia_device *p_dev->resource[0,1] for up to two ioport
  30   ranges. After a call to pcmcia_request_io(), the ports found there
  31   are reserved, after calling pcmcia_request_configuration(), they may
  32   be used.
  34* No dev_info_t, no cs_types.h (as of 2.6.36)
  35   dev_info_t and a few other typedefs are removed. No longer use them
  36   in PCMCIA device drivers. Also, do not include pcmcia/cs_types.h, as
  37   this file is gone.
  39* No dev_node_t (as of 2.6.35)
  40   There is no more need to fill out a "dev_node_t" structure.
  42* New IRQ request rules (as of 2.6.35)
  43   Instead of the old pcmcia_request_irq() interface, drivers may now
  44   choose between:
  45   - calling request_irq/free_irq directly. Use the IRQ from *p_dev->irq.
  46   - use pcmcia_request_irq(p_dev, handler_t); the PCMCIA core will
  47     clean up automatically on calls to pcmcia_disable_device() or
  48     device ejection.
  49   - drivers still not capable of IRQF_SHARED (or not telling us so) may
  50     use the deprecated pcmcia_request_exclusive_irq() for the time
  51     being; they might receive a shared IRQ nonetheless.
  53* no cs_error / CS_CHECK / CONFIG_PCMCIA_DEBUG (as of 2.6.33)
  54   Instead of the cs_error() callback or the CS_CHECK() macro, please use
  55   Linux-style checking of return values, and -- if necessary -- debug
  56   messages using "dev_dbg()" or "pr_debug()".
  58* New CIS tuple access (as of 2.6.33)
  59   Instead of pcmcia_get_{first,next}_tuple(), pcmcia_get_tuple_data() and
  60   pcmcia_parse_tuple(), a driver shall use "pcmcia_get_tuple()" if it is
  61   only interested in one (raw) tuple, or "pcmcia_loop_tuple()" if it is
  62   interested in all tuples of one type. To decode the MAC from CISTPL_FUNCE,
  63   a new helper "pcmcia_get_mac_from_cis()" was added.
  65* New configuration loop helper (as of 2.6.28)
  66   By calling pcmcia_loop_config(), a driver can iterate over all available
  67   configuration options. During a driver's probe() phase, one doesn't need
  68   to use pcmcia_get_{first,next}_tuple, pcmcia_get_tuple_data and
  69   pcmcia_parse_tuple directly in most if not all cases.
  71* New release helper (as of 2.6.17)
  72   Instead of calling pcmcia_release_{configuration,io,irq,win}, all that's
  73   necessary now is calling pcmcia_disable_device. As there is no valid
  74   reason left to call pcmcia_release_io and pcmcia_release_irq, the
  75   exports for them were removed.
  77* Unify detach and REMOVAL event code, as well as attach and INSERTION
  78  code (as of 2.6.16)
  79       void (*remove)          (struct pcmcia_device *dev);
  80       int (*probe)            (struct pcmcia_device *dev);
  82* Move suspend, resume and reset out of event handler (as of 2.6.16)
  83       int (*suspend)          (struct pcmcia_device *dev);
  84       int (*resume)           (struct pcmcia_device *dev);
  85  should be initialized in struct pcmcia_driver, and handle
  88* event handler initialization in struct pcmcia_driver (as of 2.6.13)
  89   The event handler is notified of all events, and must be initialized
  90   as the event() callback in the driver's struct pcmcia_driver.
  92* pcmcia/version.h should not be used (as of 2.6.13)
  93   This file will be removed eventually.
  95* in-kernel device<->driver matching (as of 2.6.13)
  96   PCMCIA devices and their correct drivers can now be matched in
  97   kernelspace. See 'devicetable.txt' for details.
  99* Device model integration (as of 2.6.11)
 100   A struct pcmcia_device is registered with the device model core,
 101   and can be used (e.g. for SET_NETDEV_DEV) by using
 102   handle_to_dev(client_handle_t * handle).
 104* Convert internal I/O port addresses to unsigned int (as of 2.6.11)
 105   ioaddr_t should be replaced by unsigned int in PCMCIA card drivers.
 107* irq_mask and irq_list parameters (as of 2.6.11)
 108   The irq_mask and irq_list parameters should no longer be used in
 109   PCMCIA card drivers. Instead, it is the job of the PCMCIA core to
 110   determine which IRQ should be used. Therefore, link->irq.IRQInfo2
 111   is ignored.
 113* client->PendingEvents is gone (as of 2.6.11)
 114   client->PendingEvents is no longer available.
 116* client->Attributes are gone (as of 2.6.11)
 117   client->Attributes is unused, therefore it is removed from all
 118   PCMCIA card drivers
 120* core functions no longer available (as of 2.6.11)
 121   The following functions have been removed from the kernel source
 122   because they are unused by all in-kernel drivers, and no external
 123   driver was reported to rely on them:
 124        pcmcia_get_first_region()
 125        pcmcia_get_next_region()
 126        pcmcia_modify_window()
 127        pcmcia_set_event_mask()
 128        pcmcia_get_first_window()
 129        pcmcia_get_next_window()
 131* device list iteration upon module removal (as of 2.6.10)
 132   It is no longer necessary to iterate on the driver's internal
 133   client list and call the ->detach() function upon module removal.
 135* Resource management. (as of 2.6.8)
 136   Although the PCMCIA subsystem will allocate resources for cards,
 137   it no longer marks these resources busy. This means that driver
 138   authors are now responsible for claiming your resources as per
 139   other drivers in Linux. You should use request_region() to mark
 140   your IO regions in-use, and request_mem_region() to mark your
 141   memory regions in-use. The name argument should be a pointer to
 142   your driver name. Eg, for pcnet_cs, name should point to the
 143   string "pcnet_cs".
 145* CardServices is gone
 146  CardServices() in 2.4 is just a big switch statement to call various
 147  services.  In 2.6, all of those entry points are exported and called
 148  directly (except for pcmcia_report_error(), just use cs_error() instead).
 150* struct pcmcia_driver
 151  You need to use struct pcmcia_driver and pcmcia_{un,}register_driver
 152  instead of {un,}register_pccard_driver
 153 kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.