Bug 801 - Connection breaks down with weak signal, firmware restart
: Connection breaks down with weak signal, firmware restart
Status: VERIFIED FIXED
: IPW2200
firmware error
: 1.0.6
: All All
: P2 normal
Assigned To:
:
:
:
:
:
  Show dependency treegraph
 
Reported: 2005-10-03 18:44 by
Modified: 2006-01-11 00:07 (History)


Attachments
debug log with 0x43fff (65.95 KB, application/x-gzip)
2005-10-03 18:55, Ben M Cahill
Details
Patch for ipw2200-1.0.6 (2.42 KB, patch)
2005-10-03 19:15, Ben M Cahill
Details | Diff


Note

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


Description From 2005-10-03 18:44:50
Bad performance in areas of weak reception, with firmware restarts.

From: Rolf Offermanns [rolf@offermanns.de]

My system: Maxdata 8100IS Notebook with ipw2200.
Ubuntu breezy, kernel 2.6.12-9. Self compiled, since ubuntu has debug
disabled for the ipw2200 driver.

My network setup:
AP = Draytek Vigor 2500We
WEP128
ESSID hidden

Here is the debug log.

Logs were taken while I was almost out of range. At the end of the logs I went
nearer to the AP and back again. You will see, that the connection was much
better then.  The problems don't occur when I am next to the AP.

If you need any further information, please say so. I would love to have a
working driver.

I am a developer myself, so patching, diffing, compiling, etc. are no problem.
So if you have any patches I should try, I am more than willing to help.

> Does it occur on Windows if installed?
No. Everything works fine then.

Thanks for your work,
Rolf
------- Comment #1 From 2005-10-03 18:55:38 -------
Created an attachment (id=545) [details]
debug log with 0x43fff

see above.
------- Comment #2 From 2005-10-03 19:15:29 -------
Created an attachment (id=546) [details]
Patch for ipw2200-1.0.6

The log shows an inadvertant "associate" command when the firmware was already
associated.  This was caused by a call to ipw_roam() (2nd pass) after a scan
was started, then aborted, by ipw_handle_missed_beacon(), and status flags were
in an unexpected state due to a new missed beacon response "slipping in" from
firmware at an unexpected time.

The patch cleans up the behavior of ipw_handle_missed_beacon(), and prevents
the call to ipw_roam() after an aborted scan.

To make this work better, change IPW_MB_DISASSOCIATE_THRESHOLD_DEFAULT to a
higher number, such as 24 (this will be in ipw2200-1.0.7), which will give the
scan/roam logic some time to work before giving up.

We should also double check to make sure that ipw_roam() *will* be called
(again) after a disassociate caused by 1st pass of ipw_roam() ... I haven't
seen how that works yet, but sometimes I can't find ketchup in the
refrigerator, either.

Greg reports:

Wow, that's an improvement.  It used to be that heavy VNC traffic would cause
very frequent firmware restarts.  With hwcrypto=0 it was a bit better but this
patch so far seems to have really fixed it (with hwcrypto=0 and hwcrypto=1). 

Thanks!

Cheers!greg
------- Comment #3 From 2005-10-13 07:12:53 -------
Applied for 1.0.7
------- Comment #4 From 2006-01-11 00:07:29 -------
I tested on ipw2200_1.0.10 with ieee80211_1.1.8.
I tried this case with weak reception with WEP128. There is still firmware error
on several kernel.
The firmware error will be tracked in bug#779 and other related bugs.

I didn't find other error except firmware error, so I verify this bug.