IrDA patches for 2.6.X

[Jean Tourrilhes <jt@bougret.hpl.hp.com>]

	Hi Jeff,

	Various IrDA patches. Some you have already seen, some
finishing stuff started, and a few nice critical bugs/race conditions
I spent the week tracking for a user.
        Patches tested on 2.6.0-test4, please push to Linus...

	Thanks...

	Jean

----------------------------------------------------------------------

[FEATURE] : Add a new feature to the IrDA stack
[CORRECT] : Fix to have the correct/expected behaviour
[CRITICA] : Fix potential kernel crash

ir2604_ircomm_owner-2.diff :
~~~~~~~~~~~~~~~~~~~~~~~~~~
	o [FEATURE] Finish removing traces of old module refcount stuff

ir260_nsc_39x_fixes.diff :
~~~~~~~~~~~~~~~~~~~~~~~~
		<Patch from Jan Frey>
	o [CORRECT] Make NSC 3839x probe and init *really* work
		The new 3839x code was totally broken.
		Won't affect code for regular 38108/38338 chips.

ir2604_irtty_cleanup-2.diff :
~~~~~~~~~~~~~~~~~~~~~~~~~~~
	o [FEATURE] Finish removing traces of old irtty driver

ir2604_lap_close_race-5.diff :
~~~~~~~~~~~~~~~~~~~~~~~~~~~~
	o [CRITICA] Fix a race condition when closing the LAP
		prevent the stack to open new LSAPs while we are killing them.

ir2604_connect_watchdog-5.diff :
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
	o [CRITICA] In case of connect watchdog, drop reference to the LAP
	o [CORRECT] Prevent dumping LSAP after connect watchdog
	o [CRITICA] Prevent dumping TSAP if dumping LSAP did fail
	o [CORRECT] Only set connected bit on response if LSAP state is correct

ir2604_init_cleanup-6.diff :
~~~~~~~~~~~~~~~~~~~~~~~~~~
		<Patch from Guennadi Liakhovetski>
	o [FEATURE] Don't leak stuff in various failure paths
	o [FEATURE] Properly initialise self->max_header_size in IrIAP

ir2604_modules_aliases.diff :
~~~~~~~~~~~~~~~~~~~~~~~~~~~
		<Patch from Rusty Russell>
	o [FEATURE] Add module aliases to dongle drivers