Prism2/2.5/3 Host AP - new stable release v0.2.5 - 2004-10-03

[Jouni Malinen]

A new version of Prism2/2.5/3 Host AP was just released and is now
available from http://hostap.epitest.fi/

This release is a bug fix version from the stable (0.2.x) branch and
replaces 0.2.4 as the recommended stable version. However, people using
Linux 2.2.x versions or older 2.4.x/2.6.x versions may have to use the
old stable series (0.1.x) since the 0.2.x versions do not support some
of the older kernel versions. Main development continues in the
development branch (0.3.x).


hostap-driver:
* fixed card enabling after firmware download in case any of the
  netdevs were up when the download was started
* fixed netif_carrier_on/off() calls to leave carrier on for Master
  mode; previously this may have been left off in some cases which
  could prevent packet bridging with new kernel versions

hostapd:
* fixed EAPOL reauthentication to trigger WPA/WPA2 reauthentication
* fixed EAPOL state machine to stop if STA is removed during
  eapol_sm_step(); this fixes at least one segfault triggering bug with
  IEEE 802.11i pre-authentication
* fixed an alignment issue that could cause SHA-1 to fail on some
  platforms (e.g., Intel ixp425 with a compiler that does not 32-bit
  align variables)
* fixed RADIUS reconnection after an error in sending interim
  accounting packets

wpa_supplicant:
* wpa_cli: fixed parsing of -p <path> command line argument
* fixed parsing of wep_tx_keyidx
* fixed couple of errors in PCSC handling that could have caused
  random-looking errors for EAP-SIM
* PEAPv1: fixed tunneled EAP-Success reply handling to reply with TLS
  ACK, not tunneled EAP-Success (of which only the first byte was
  actually send due to a bug in previous code); this seems to
  interoperate with most RADIUS servers that implements PEAPv1
* PEAPv1: added support for terminating PEAP authentication on tunneled
  EAP-Success message; this can be configured by adding
  peap_outer_success=0 on phase1 parameters in wpa_supplicant.conf
  (some RADIUS servers require this whereas others require a tunneled
  reply
* PEAPv1: changed phase1 option peaplabel to use default to 0, i.e., to
  the old label for key derivation; previously, the default was 1,
  but it looks like most existing PEAPv1 implementations use the old
  label which is thus more suitable default option
* changed SSID configuration in driver_wext.c (used by many driver
  interfaces) to use ssid_len+1 as the length for SSID since some Linux
  drivers expect this
* fixed couple of unaligned reads in scan result parsing to fix WPA
  connection on some platforms (e.g., ARM)

hostap-utils:
* no changes - 0.2.5 skipped


-- 
Jouni Malinen                                            PGP id EFC895FA