Upon start-up netcfg attempts to bring a bridge interface up for hostapd to use as an Access Point. However, part of the netcfg scripts is detecting whether there is a signal and this causes the bridge interface to be set down on failure. This was working in prior versions to 2.6.6. It may also be a kernel update where no-carrier is being returned.

/etc/network.d/bridge:
INTERFACE="br0"
CONNECTION="bridge"
DESCRIPTION="Bridge for wireless"
IP="static"
ADDR="192.168.1.1"
NETMASK="255.255.255.0"

script code that is causing the interface to be set down
/usr/lib/network/connections/ethernet:
if ! checkyesno "${SKIPNOCARRIER:-no}" && ip link show dev "$INTERFACE" | fgrep -q "NO-CARRIER"; then
sleep ${CARRIER_TIMEOUT:-2} # Some cards are plain slow to come up. Don't fail immediately.
# if ip link show dev "$INTERFACE" | fgrep -q "NO-CARRIER"; then
# report_iproute "No connection"
# fi
fi

I have commented out the test to keep the interface from being brought down on boot. Maybe there needs to be a flag that an empty bridge interface is legitimate and will not have a carrier immediately (ie skip the test)?