#8667 HIGH 8.2.1: wpa_supplicant generates many redundant scan requests before associating

Zarro Boogs per Child bugtracker at laptop.org
Mon Sep 29 20:31:33 EDT 2008


#8667: wpa_supplicant generates many redundant scan requests before associating
------------------------+---------------------------------------------------
   Reporter:  jcardona  |       Owner:  dsaxena      
       Type:  defect    |      Status:  new          
   Priority:  high      |   Milestone:  8.2.1        
  Component:  wireless  |     Version:  not specified
 Resolution:            |    Keywords:               
Next_action:  package   |    Verified:  0            
  Blockedby:            |    Blocking:               
------------------------+---------------------------------------------------

Comment(by carrano):

 Repeated the association and scanning tests again with the
 [http://dev.laptop.org/~dsaxena/kernel-2.6.25-20080926.1.olpc.5f842a369c6d018.i586.rpm
 latest test kernel], but this time with the
 [http://dev.laptop.org/attachment/ticket/8666/usb8388_5_110_22_p255.tar.gz
 test firmware]

 I am attaching fail.Harumaki, which is a dmesg output (scanning and
 association debug on) for a failed association to SSID Harumaki (WPA AP)

 From a user perspective, association hits seems more frequent now, but it
 is really hard to conclude anything from this observation now, since the
 variance in all the tests so far are still very high.

 Also repeated the tests in which a scanning is performed immediately after
 a failed association. The only debug flag on was scan.

 {{{
 dmesg -c; sleep 1; iwlist eth0 scan; sleep 20; dmesg | tee <output-file>
 }}}

 Scanning ok, after a failed association: scan.22p255.ok.1

 Scanning fails, after a failed association: scan.22p255.fail.1

 Mesh disabled, scanning ok, after a failed association:
 scan.22p255.nomesh.ok.1

 Mesh disabled, scanning ok, after a failed association: curiously dmesg
 register no output during this 20 seconds in this case (repeated this
 three times)

 No firmware "events" were registered in any of the above tests.

-- 
Ticket URL: <http://dev.laptop.org/ticket/8667#comment:16>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system


More information about the Bugs mailing list