always network association failed about iwidarwin HOT 14 CLOSED

uhhm. 

"AUTH_SEQ_1" always is returned after sending ipw_send_cmnd related with
ipw_associate_network.

and disassociation.

some networks use AUTH_SEQ_1 but after a while they skip it and do the 
association.
that's ok.

does that network goes to the exclude list?

after reboot i always get the network excluded. i've to restart the adapter to
reassociate. we should do a delay? (wait for mac os to boot)

my pc try to be associating the network for 1 hours , but cannot connect .
and always "AUTH_SEQ_1".

getting "AUTH_SEQ_1" and "disassociate: " per calling ipw_associate.
my pc iterate ipw_scan and ipw_associate and "dissociate:",every.
I dont have experience to connnet in current trunk.

I see the same post in forum. all 2934abg may not have availability in this 
trunk.

i find the problem

the ipw_handle_data_pkg meed to be fix. i try several changes and get speed 
inprovements but it always crash 
after a while.

i've also comment ieee80211_rx to use only the needed code

pleasy try to fix these

we're almost done with rx

:)

waht important detail:

my card as a wireless button. i used it to power on off the card

how do you do this with your card? do you use keyboard fn+fx?

if you can't turn the card on/off the ip filter will exclude you network and 
won't allow to associate!

> my card as a wireless button. i used it to power on off the card
> how do you do this with your card? do you use keyboard fn+fx?

wireless button which you say is  hardware button?

there are not wireless button  in my machine.
2945 can turn on/off in linux  and windows by software.
( in linux via /proc or /sys )


> if you can't turn the card on/off the ip filter will exclude you network and 
won't
allow to associate!
>

if this is true, you must remove routine of the ip filter or chage .

> the ipw_handle_data_pkg meed to be fix.
> i try several changes and get speed inprovements but it always crash 
> after a while.

why mbuf_setlen is called?
the problem of my older code  is no-set correctly mbuf_len .

i seem that always mbuf_len = 0 .
plz add the code after setting mbuf_pkthdr_setlen.


IWI_DEBUG("dump rx packet size mbuf_len = %d,mbuf_pkthdr_len %d , rx frame size 
%d.
mbuf_next is %s\n",mbuf_len(rxb->skb),mbuf_pkthdr_len(rxb->skb),
mbuf_next(rxb->skb) ? "exist"  : "nothing");


if mbuf_len != mbuf_pkthdr_len  and "mbuf_next is exist",
mbuf_setlen(le16_to_cpu(pkt->u.frame.length)) is required.

i wil disable ip filter for now. but you also have this problem after reboot if 
p_disable=0 you get an invalid ip 
and the network never associate - this happen with my card!!

this also happen if the link goes down or get a week signal

i solve this by pressing the wireless button off and then on

i've seen this problem on a place when i can get 2 networks:

- the 1 has a stronger signal but invalid ip (169.254.xxx)
- the 2 has a week signal and never get associated because the driver always 
choose the net with a better 
signal

this is the reason why i made the ip filter (it excludes the 1 net and 
associate to the 2nd)



how can we solve this for your card?

one solution is to do a delay to allow mac os x to finish booting
 but how to fix for the case of having 2 networks?

move to rx performance issue about mbuf_len.



this problem( correctly cannot get correctly ip address) occurs after merge ur 
new
feature. (perhaps in revision 119).
so i used static IP.( association is work , no problem).

association in current verision is not work. 
and i got "parity error".

> - the 1 has a stronger signal but invalid ip (169.254.xxx)

our driver may get no-correctly ionetworkinterface.
in other mac machine.

% netstat -nr
169.254            link#4             UCS         0        0    en0
192.168.0          link#4             UCS         4        0    en0
192.168.0.1        link#4             UHLW        5        0    en0
192.168.0.15       127.0.0.1          UHS         0        4    lo0

and en0 is ethernet interface. 
we research about this. 

> - the 2 has a week signal and never get associated 
> because the driver always choose the net with a better

this is correct as specification.
if there are the strong signal network and the weak signal network,
do you want to set weak signal network ,automaticaly?

we should apply its function about user specified,now.

we should implement it with implementing setting via airport menu  after sevaral
fixed in future.

i have a solution for your card:

what about changing the code of get/set POWER()

if you can do this you can use the airport menu to turn the card on/off

this will be a good fix for all users of all drivers

>our driver may get no-correctly ionetworkinterface.
>in other mac machine.

i've checked it in windows and the 169.254.xxx network did have a better signal 
but the 2 network (worst 
signal) was the one who gives a valid ip and allow connect.

of course in windows you can choose the network...

either way i disabled ip filter in svn. i keep the log as warning

why did you change ieee80211.h? i need the int exclude; there

kazu,

what about this (from  ipw2200.c)

static inline int ipw_alive(struct ipw_priv *priv)
{
    return ipw_read32(priv, 0x90) == 0xd55555d5;
}

can this be used to turn the card on/off in the get/setpower()?

if not how can we find out a way to do this by software?

?????

this problem dont occur in v6. why is it? 

> i've checked it in windows and the 169.254.xxx 
> network did have a better signal but the 2 network (worst
> signal) was the one who gives a valid ip and allow connect.

I may not understand 169.254.xxx network which you say.
plz say strictly word.

> of course in windows you can choose the network...

sorry, plz all information of 169.254.xxx network.

> why did you change ieee80211.h? i need the int exclude; there

you didnt commit changing of ieee80211.h when you commit your new feature.
so i commited to add "int exclude;"  in rev 156.

meaning of "* fixed a part of no-merging." is 
"jalavoui commited new feature but 
 changing of ieee80211.h didnt be commited 
  so i added line of 'int exclude' and  commited it" 

do you undersntad?

% cd trunk/iwi2200
% LANG=C svn log -r153:156 
------------------------------------------------------------------------
r153 | jalavoui | 2007-02-16 17:35:08 +0100 (Fri, 16 Feb 2007) | 1 line

freepacket fix
------------------------------------------------------------------------
r154 | jalavoui | 2007-02-16 17:50:20 +0100 (Fri, 16 Feb 2007) | 1 line

lock removing - this fixes kernel panic!
------------------------------------------------------------------------
r156 | kazuhisa | 2007-02-17 09:54:43 +0100 (Sat, 17 Feb 2007) | 1 line

* fixed a part of no-merging.
------------------------------------------------------------------------

% LANG=C svn diff -r155:156   net/ieee80211.h | less
Index: net/ieee80211.h
===================================================================
--- net/ieee80211.h     (revision 155)
+++ net/ieee80211.h     (revision 156)
@@ -1224,7 +1224,8 @@
        /* Ensure null-terminated for any debug msgs */
        u8 ssid[IW_ESSID_MAX_SIZE + 1];
        u8 ssid_len;
-
+       int exclude;
+       
        struct ieee80211_qos_data qos_data;

        /* These are network statistics */

I turn off with setting "1" /sys/... rf_kill.
and i turn on with settig "0" /sys/... rf_kill.
I think that it is call ipw_rf_kill.

i seem that power on/off is dont related with my disassociation problem.
following is my log.

iwi2200:  ipw_associate() associate...
iwi2200:  ipw_best_network() ipw_best_network
iwi2200:  ipw_best_network() iw_mode[0] capability[1] flag[2] scan_age[1500]
iwi2200: ipw_best_network() Network 'vantronix (00:90:cc:46:28:75)'
iwi2200: Network 'vantronix (00:90:cc:46:28:75)' is a viable match.
iwi2200: ipw_best_network() ipw_best_network
....
ipw_associate() trying to associate 'vantronix (00:90:cc:46:28:75)'
ipw_associate_network() Assocation attempt: 'vantronix', channel 34, 802.11a 
[8],
long[:long], enc=off.
iwi2200:  ipw_associate_network() SSID command.
iwi2200:  __ipw_send_cmd() SSID command (#8) 9 bytes: 0x00000023
iwi2200:  handleInterrupt() Command completed.
iwi2200:  ipw_associate_network() IPW_CMD_SUPPORTED_RATES .
iwi2200:  __ipw_send_cmd() SUPPORTED_RATES command (#22) 16 bytes: 0x00000023
iwi2200:  handleInterrupt() Command completed.
iwi2200:  ipw_associate_network() sys config command.
iwi2200:  __ipw_send_cmd() SYSTEM_CONFIG command (#6) 20 bytes: 0x00000023
iwi2200:  ipw_associate_network() Association sensitivity: -67
iwi2200:  handleInterrupt() Command completed.
iwi2200:  __ipw_send_cmd() SENSITIVITY_CALIB command (#42) 4 bytes: 0x00000023
iwi2200:  handleInterrupt() Command completed.
iwi2200:  __ipw_send_cmd() ASSOCIATE command (#21) 40 bytes: 0x00000023
iwi2200:  handleInterrupt() Command completed.
iwi2200:  ipw_associate_network() associating: 'vantronix' 00:90:cc:46:28:75 
iwi2200:  ieee80211_rx_mgt() (80211_MGMT) received BEACON (128)
iwi2200:  ieee80211_process_probe_response() (80211_SCAN) Updating 'vantronix'
(00:90:cc:46:28:75) via BEACON.
iwi2200:  handleInterrupt() Command completed.
iwi2200:  __ipw_send_cmd() ASSOCIATE command (#21) 40 bytes: 0x00000023
iwi2200:  handleInterrupt() Command completed.
iwi2200:  ipw_associate_network() associating: 'vantronix' 00:90:cc:46:28:75 
iwi2200:  handleInterrupt() IPW_INTA_BIT_RX_TRANSFER)
iwi2200:  notifIntr() type = 11 (4 bytes)
iwi2200:  notifIntr() AUTH_SEQ_1
iwi2200:  notifIntr() type = 10 (4 bytes)
iwi2200: disassociated: 'vantronix' 00:90:cc:46:28:75 
....
following these is interated ever.

I found why the problem is caused by no-seting priv->mac_addr.

when getHarwareAddress is called
if priv is not initialized, priv->mac_addr is not set.

this is fixed in r164.