Skip to contentSkip to conversation

PCMCIA patches

CFT Russell King 1 participant 1 message Jun 23, 2003

lore
Share & tools
lore
#0 lore raw

Ok guys,

Here's another set of PCMCIA patches to keep people occupied for a while. Tested sa11xx and yenta here. Please report successes/failures.

Note: if you are using modules, you will only be able to remove the socket driver when the cards are ejected (by either physically removing the card or via cardctl eject.) Some init scripts may get upset with this on shutdown; this will eventually be noted in davej's 2.6 changes document. The script needs to run cardctl eject before rmmoding the pcmcia modules.

 drivers/pcmcia/cs.c           |  103 ++++++++++++++++++++++++------------------
 drivers/pcmcia/i82092.c       |   46 ++----------------
 drivers/pcmcia/i82092aa.h     |    1
 drivers/pcmcia/i82365.c       |   75 +++---------------------------
 drivers/pcmcia/sa11xx_core.c  |   66 +++++---------------------
 drivers/pcmcia/sa11xx_core.h  |    3 -
 drivers/pcmcia/tcic.c         |   48 ++-----------------
 drivers/pcmcia/yenta_socket.c |   39 +--------------
 drivers/pcmcia/yenta_socket.h |    5 --
 include/pcmcia/ss.h           |    3 -
 10 files changed, 96 insertions(+), 293 deletions(-)

http://patches.arm.linux.org.uk/pcmcia/pcmcia-event-20030623-1.diff

Move ->owner field from socket operations to pcmcia_socket. (This change is mainly for the SA11xx drivers, which use a core driver for the chip, and a separate module for all the machine specific bits.)

http://patches.arm.linux.org.uk/pcmcia/pcmcia-event-20030623-2.diff

Get/Put module when we insert and remove a card. This avoids a potential deadlock when socket drivers are unloaded, and we have a cardbus card known to the system.

http://patches.arm.linux.org.uk/pcmcia/pcmcia-event-20030623-3.diff

Remove original module use accounting in register_callback.

http://patches.arm.linux.org.uk/pcmcia/pcmcia-event-20030623-4.diff

Add work-around for i82365-based socket drivers to the core PCMCIA code. Since insert processing is not a time critical event, we can afford to delay (by sleeping) these for everyone.

http://patches.arm.linux.org.uk/pcmcia/pcmcia-event-20030623-5.diff

Remove register_callback methods; allow socket drivers to call pcmcia_parse_events() directly.

http://patches.arm.linux.org.uk/pcmcia/pcmcia-event-20030623-6.diff

Remove now obsolete work queues, spinlocks, and code from socket drivers.

Signature · 2 lines
Russell King ([email protected])                The developer of ARM Linux
             http://www.arm.linux.org.uk/personal/aboutme.html