Bug 572 - Cannot associate with AP (disassociates after AUTH_SEQ_1)
: Cannot associate with AP (disassociates after AUTH_SEQ_1)
Status: VERIFIED FIXED
: IPW2200
__UNSPECIFIED__
: 1.0.3
: IBM Gentoo
: P2 normal
Assigned To:
:
:
:
:
:
  Show dependency treegraph
 
Reported: 2005-02-08 06:04 by
Modified: 2005-10-06 15:59 (History)


Attachments
Kernel log with debug mask = 255 (13.54 KB, text/plain)
2005-02-08 06:05, Florian Hackenberger
Details
kernel log with debug = 255 (97.43 KB, text/plain)
2005-05-10 02:53, Benoit Boissinot
Details
fix iwconfig txpower auto (566 bytes, patch)
2005-05-11 01:24, Zhu Yi
Details | Diff
a better fix (1.65 KB, patch)
2005-05-11 23:47, Zhu Yi
Details | Diff
Correctly merges chuyee's patch into 1.0.4 vs. what actually made it into the build (495 bytes, patch)
2005-05-19 07:41, James Ketrenos
Details | Diff


Note

You need to log in before you can comment on or make changes to this bug.


Description From 2005-02-08 06:04:09
I cannot associate with the AP at my university. It runs at 801.11b and has 
quite a low signal strength (35 reported by windows driver in ndiswrapper). I 
have attached the logs. As you can see it disassociates right after recieving a 
notification with AUTH_SEQ_1 set. But it does work with the ndiswrapper, so it 
has to be either a firmware or a driver problem. There was a thread on the 
mailing list discussing the problem, but there was just one response from the 
developers, which did not resolve the problem. I classified this bug as 
critical, because it prevents me from using the wireless network, which is the 
purpose of the driver.
------- Comment #1 From 2005-02-08 06:05:51 -------
Created an attachment (id=264) [details]
Kernel log with debug mask = 255
------- Comment #2 From 2005-02-10 06:20:51 -------
I have found the cause of the error. It is not the ipw2200 driver, but the 
Gentoo specific wireless scripts, which "somehow" cause this behaviour. 
------- Comment #3 From 2005-04-28 09:38:44 -------
could it be reopened ?

i narrowed the bug, and it is ipw2200 fault (not gentoo's scripts).

if you execute the following commands:

iwconfig eth1 txpower auto
(the time between the 2commands doesn't matter)
iwconfig eth1 mode auto

the card cannot associate anymore (with the symptoms described in the previous
comments - AUTH_SEQ_1).

It is interesting to note that when you run the two commands in the reverse
order but it as to be quick (like command1 ; command2), the card will fail too.


Hope it helps,

Benoit
------- Comment #4 From 2005-05-10 01:43:12 -------
Does the problem still exist on 1.0.3?

If 1.0.3 still has the problem, please try the patch in bug 455.
http://www.bughost.org/bugzilla/attachment.cgi?id=341&action=view
------- Comment #5 From 2005-05-10 01:51:56 -------
For me the bug is resolved in 1.0.3, I'll leave it up to you to close the bug. 
------- Comment #6 From 2005-05-10 01:55:38 -------
1.0.3 fixes this
------- Comment #7 From 2005-05-10 01:57:29 -------
1.0.3 fixes this
------- Comment #8 From 2005-05-10 01:57:52 -------
fixed in 1.0.3
------- Comment #9 From 2005-05-10 02:21:31 -------
I am having the problem with 1.0.3 (gentoo, Dell D600), and the patch does not
fix it.

I can provide more info if needed.

regards,

Benoit.
------- Comment #10 From 2005-05-10 02:43:02 -------
Benoit, please provide dmesg with debug=255 for the patched 1.0.3 driver.
------- Comment #11 From 2005-05-10 02:53:11 -------
Created an attachment (id=371) [details]
kernel log with debug = 255

Kernel log while issuing the commands :
iwconfig eth1 txpower auto
iwconfig eth1 mode auto
------- Comment #12 From 2005-05-11 01:24:32 -------
Created an attachment (id=374) [details]
fix iwconfig txpower auto

iwconfig will pass -1 if txpower auto is used, this decrease the tx power for
the device. Please try this patch and see if it fixes the bug.
------- Comment #13 From 2005-05-11 01:43:33 -------
It fixes it (with or without the patch in bug #455).

Thanks,

Benoit
------- Comment #14 From 2005-05-11 02:12:15 -------
mark as fixed
------- Comment #15 From 2005-05-11 23:47:28 -------
Created an attachment (id=377) [details]
a better fix

I'll check in this fix to next release.
------- Comment #16 From 2005-05-19 04:51:15 -------
The patch is in 1.0.4. 
------- Comment #17 From 2005-05-19 07:37:07 -------
*** Bug 632 has been marked as a duplicate of this bug. ***
------- Comment #18 From 2005-05-19 07:41:26 -------
Created an attachment (id=390) [details]
Correctly merges chuyee's patch into 1.0.4 vs. what actually made it into the
build

In addition to fixing what this bug was originally about, chuyee's patch also
fixes support for 'txpower auto' to actually set the power level to a default,
which allows us to support that parameter as well (fixing #632)
------- Comment #19 From 2005-05-19 09:01:58 -------
*** Bug 668 has been marked as a duplicate of this bug. ***
------- Comment #20 From 2005-05-19 09:23:15 -------
There seem to be several different interpretations of what settings txpower to
auto should do.
a) from iwconfig(8) in wireless-tools-27:
   In  addition,  on and off enable and disable the radio, and auto
   and fixed enable and disable power control  (if  those  features
   are available).

b) chuyee's patch:
   choose a reasonable default

c) Planet WL3501 driver in my 2.6.11 (gentoo) source tree reports fixed=0,
because txpower is configurable (not fixed).

Other drivers in the kernel tree don't support txpower, can't change it, or
simply ignore fixed.  I think we need some clarification as to which
interpretation is correct (after all, wireless extensions should be consistent
across drivers).