aboutsummaryrefslogtreecommitdiffstats
path: root/todo.lst
blob: 80e802e2353863e1d1e9b4c90b5b5f046970f8bc (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
- WDS:
  * statistics for wlan#wds# interfaces (fix TX packet part of dev->get_stats,
    add dev->wireless_stats)
- USB support
  * there is a patch for old Host AP driver version; merge it into current
    version
- IAPP (started, but not yet complete)
- Mobile IP
- test iwe_stream_add_*() functions with different gcc version
  (it looks like following code is executed in wrong order:
       iwe->len = event_len;
       memcpy(stream, (char *) iwe, event_len);
- allow more STA info to be sent for iwspy with WIRELESS_EXT >= 15
  (max num increased for get, but set is still limited to 8)
- Makefile improvements:
  * add option for reading configuration (SMP, modversions, omit_fp) from
    running kernel
  * warn if configuration from source directory and running kernel do not match
    (i.e., if the compiled hostap_*.o would not work with running kernel)
- debug prism2_tx_timeout() reseting / COR sreset during high load; did not
  seem to be able to reset PLX card properly (card needed host reboot)
- better support for different wlan hw:
  * hostapd should already work with other devices
  * hostap_ap.c should be separated into its own module and the interface from
    the low-level driver should use IEEE 802.11 headers without Prism2 specific
    data (if HW specific data is needed, it could be added to the end of the
    skb buffer)
  * maybe test with Cisco 340 (airo_cs.o and some minimal changes for AP mode)
- use TX callback in hostapd to make sure that EAPOL-Key frames are received
  before setting new WEP key; resend EAPOL-Key frame couple of times,
  log event; have to change keys anyway
- figure out better way of indicating TX callback request/reply between kernel
  driver and hostapd (using 802.11 header protocol field is not exactly nice..)
- implement more intelligent TX rate controlling
- remove expired APs more quickly
- make sure that hostscan setting is re-set after card reset
- ioctl's can be interrupted (EINTR, e.g., by killing iwconfig) which can
  result in leaving entries to pending CmdCompl list or command completion
  events to empty pending list; this should be cleaned up
- try using userland helper to reload firmware upgrade into RAM on card reset
  (see, e.g., hotplug subsystem and /sbin/hotplug)
- check wireless_copy.h and different ioctl numbering for different
  wireless_ext versions
- command line arg vs. SIGHUP (now debug/daemonize will change on HUP)
- FIX: IAPP:
   Received 16 byte IAPP frame from 127.0.0.1
   IAPP: version=0 command=0 id=0 len=16
   Received IAPP-ADD for STA 00:50:8b:46:94:52 (seq# 484) from 127.0.0.1:2313
   Removing STA 00:50:8b:46:94:52 due to IAPP-ADD notification from 127.0.0.1
   IEEE 802.1X: station 00:50:8b:46:94:52 port disabled
- RX: could possibly remove one tasklet and use dev_queue_xmit() directly
  since it is now in softirq context
- add option for making wlan#wds? interfaces point-to-point
  * is '/sbin/ip ad ad 192.168.1.1 peer 192.168.5.1 dev eth0' enough?
# /sbin/ip ad
2: eth0: <BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast qlen 100
    link/ether 00:06:5b:a2:b3:8a brd ff:ff:ff:ff:ff:ff
    inet 192.168.1.1 peer 192.168.5.1/32 scope global eth0
- deauth if WEP frame received when there is no key for the used keyid
- STA f/w 1.4.9 does not seem to provide beacon frames to the driver in
  Host AP mode --> automatic WDS creation does not work
- Repeater mode does not send beacons -> automatic WDS link creation does not
  work
  ==> add own "beacon" for this, e.g., sending 4-addr WDS frame to broadcast
	addr every 10 sec; other APs can then use this as an indication of a
	link; include SSID in that "beacon" frame
- add SuppRates configuration into hostapd
- dBm conversions
  * comms_quality (FD43) seems to be more or less OK
  * RX desc handling does not use the documented range at least with latest
    STA firmware versions; signal/silence range seems to be 0 .. 100, not
    27 .. 154;
- problem with sta->users counter:
	STA=00:90:4b:0d:fd:89
	users=-1641
  (this happened when testing iwconfig rate command on both STA and AP;
  in addition, lot of Auth frames were not ACKed)
- add support for setting static broadcast WEP key for IEEE 802.1X
- Xsupplicant and no RFC1042 header issues(?)
  "Data frame with no RFC1042 header"
  (couple of reports on hostap and xsupplicant mailing lists)
- verify that bcrx_sta_key is documented and implemented correctly;
  individual RA address means the next recipient, not necessarily the final
  destination (DA); for example, in STA broadcast RA is BSSID, i.e., unicast
  not broadcast; thus, AP must decrypt using individual key
- rename procfs directory if interface name is changed (?)
- RX: use skb_copy() instead of skb_clone() when sending broadcast frames both
  to wireless media and Linux net stack;
  use skb_share_check() and skb_unshare()
- implement TX power controlling with feedback from TX power measurements
  BBP:
  CR31: manual TX power: 7-bit value, -64 .. +63
- move few remaining hardware independent parts from hostap_hw.c into
  hostap.c (net dev registration, etc.)
- divide hardware independent parts of hostap.ko module into a separate
  IEEE 802.11 module; remaining part will be generic Prism2/2.5/3 module
  that will be used by hostap_{cs,pci,plx} modules
- move the only remaining Prism2 specific part of hostap_ap.c, signal quality
  conversion, into hostap_hw.c that hostap_ap.c can become completely hardware
  independent
- prioritize management frames in TX side (e.g., try assoc during high
  network load situation)
- investigate whether IEEE 802.11 fragmentation could be done in the driver
  (at least when using WEP/TKIP/CCMP)
- test with more than two STAs and heavy traffic load (kernel dump/hang
  reported with an old driver version, mailing list 2003-02-12); e.g.,
  simulate 1000 STAs..
- verify that IEEE 802.1X reAuth works with WinXP Supplicant
- verify that WDS links work when bridged together with eth0
  and br0 takes the MAC address of eth0, not wlan0
- prism2_srec flash dump (in a format that could then be written back
  to the card)
- add more wireless events for iwevent (e.g., STA auth failure as text string)
- do not send multicast data frames to the wlan0 interface if there are
  no associated STAs (WDS links will still need to get these)
- verify WDS operation with WEP (some problems reported)
- verify that WDS still works in repeater mode
- fix kernel crash when running tcpdump on wifi0 when using hostapd
- fix hostapd SIGHUP config reloading to work with all configuration changes
- verify TIM update for PS mode; beacons did not seem to always have TIM
  flag for a STA even though handle_set_tim_queue() had updated cnfTimCtrl
- check Linux bridge patch at
  http://www.czfree.net/forum/showthread.php?s=f79b715a1482a1a1fa2d9f3f2ffad620&threadid=4154
  (can it be used with ap_bridge_packets=0 in all cases?; it is supposed
  to overwrite dst addr to be wlan card addr so that frames are sent
  back to the same iface)
- debug power saving mode in STA modes; RIDs are set based on iwconfig
  configuration (but this is lost on hw reset), but the card does not really
  use PS mode (except with latest STA f/w..)
- verify power saving configuration (power timeout and power period
  not showing the same values that were configured ?)
- make sure modversions works in hostap.o module
- eloop.c: add eloop_unregister_signal()
- driver/modules/Makefile overwrites pcmcia-cs Makefile when copied with
  cp -a.. this should not be done when using external pcmcia-cs package
- hostapd: MPPE key decrypt: verify that MSB of salt is set and that the Recv
  and Send keys use different salt (RFC2548)
- verify that the associating STA supports all configured basic rates (both
  driver and hostapd); reject association if any of the basic rates is not
  supported


verify:
- can pairwise TKIP key be used if default key is not set?
- are signal quality numbers updated with Linux 2.6.x?
- run scan more often in wpa_supplicant
- try to get wpa_supplicant/WPA working in host_roaming=1 because
  host_roaming=2 seems to be a bit unstable..
- does enh_sec=3 remove SSID also from ProbeResponse; this would break
  wpa_supplicant scan_ssid with Host AP driver


Linux 2.6:
- MODULE_PARM() -> module_param()
- use USER_HZ with 2.6.x when exporting jiffies to user space (procfs, etc.)