into nuts.ninka.net:/home/davem/src/BK/sctp-2.5

Fix for action bits regressed HOST_NAME_ADDRESS parm behavior, so
just special case this so the association is aborted as expected. 

into nuts.ninka.net:/home/davem/src/BK/net-2.5

already being unhashed on entry.

as we delete the entry, we can only poison the back pointer, not the
traversal pointer (rcu traversal only ever walks forward).

Make __d_drop() take this into account.

into nuts.ninka.net:/home/davem/src/BK/net-2.5

Based upon a patch from Stephen Hemminger.

into home.transmeta.com:/home/torvalds/v2.5/linux

drivers/char/pty.c converted to dynamic allocation

drivers/char/rio/rio_linux.c converted to dynamic allocation

drivers/char/epca.c converted to dynamic allocation

arch/v850/kernel/{memcons,simcons}.c converted to dynamic allocation

drivers/char/vt.c converted to dynamic allocation

arch/um/drivers/* converted to dynamic allocation

arch/cris/drivers/serial.c converted to dynamic allocation

arch/alpha/kernel/srmcons.c converted to dynamic allocation

drivers/s390/char/con3215.c converted to dynamic allocation

drivers/s390/char/sclp_tty.c converted to dynamic allocation