Bug 820167

Summary: scanning failure after doing wireless roaming on iwl6205
Product: Red Hat Enterprise Linux 6 Reporter: Stanislaw Gruszka <sgruszka>
Component: kernelAssignee: Stanislaw Gruszka <sgruszka>
Status: CLOSED NOTABUG QA Contact: Desktop QE <desktop-qa-list>
Severity: urgent Docs Contact:
Priority: urgent    
Version: 6.2CC: brunor, hesemeyt, kzhang, linville, tpelka, walicki
Target Milestone: rc   
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: 814674 Environment:
Last Closed: 2012-05-15 10:17:11 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Bug Depends On: 766952, 808095, 814674    
Bug Blocks:    
Attachments:
Description Flags
test-roam2
none
Everything in one file
none
wpa_supplicant.log files from today and yesterday none

Description Stanislaw Gruszka 2012-05-09 10:13:47 UTC
Steps to reproduce:
sudo service NetworkManager stop
sudo killall wpa_supplicant
sudo /usr/sbin/wpa_supplicant -c /etc/wpa_supplicant/wpa_supplicant.conf
-B -u -dddt -i wlan0 -D nl80211 -f /var/log/wpa_supplicant.log

Run roaming testing script.

Results:
After a while device is unable to scan, and roaming is not performed, only one AP is visible.

iwlist wlan0 scan
wlan0     Failed to read scan data : Resource temporarily unavailable

Additional info:
This happen on laptop with iwl 6205, problem is not reproducible on laptop with 5100 adapter.

Comment 1 Stanislaw Gruszka 2012-05-09 10:15:43 UTC
Created attachment 583207 [details]
test-roam2

Roaming testing script.

Comment 2 Stanislaw Gruszka 2012-05-09 10:19:35 UTC
Issue is not reproducible on RHEL6.3 beta kernel - Thorsten, is that correct?

Comment 4 Thorsten Hesemeyer 2012-05-09 11:51:06 UTC
Created attachment 583235 [details]
Everything in one file

Hi,

I've put together in one tgz file:
  test-roam2   ... script used (just to be complete)
  dmesg.txt    ... output of dmesg command
  sosreport*   ... files from sosreport of my T420
  (other)      ... ifconfig -a and some more output
The folder var/log/ contains log files from /var/log/
  dmesg        ... /var/log/dmesg file
  messages     ... /var/log/messages file
  wpa_supplicant.log ... /var/log/wpa_supplicant.log file with timestamps and debugging information
  test-roam.log... /var/log/test-roam.log file written by the script above.

The test-roam.log file contains the following fields:
  #1 number of access points found while scanning 
  #2 Date&Time /human readable,
  #3 Time in seconds since 1970
  #4 Uptime Idle in seconds
  #5 Uptime in seconds
Meant to make it easier to join and find the different time stamps from dmesg, messages and other log files.

Kind regards,
Thorsten

Comment 5 Thorsten Hesemeyer 2012-05-09 11:56:27 UTC
Hi,

the test-roam.log file gives an overview what happened:
  a) started 05/08/12 08:21:14
  b) after 05/08/12 10:26:42 only and exactly one access point visible
  c) after 05/08/12 21:49:45 no access point visible

I was astonished to see there was no access point left over this morning.
But I guess that is really good for finding the issue.

Kind regards,
Thorsten

Comment 6 Thorsten Hesemeyer 2012-05-09 12:19:50 UTC
Created attachment 583258 [details]
wpa_supplicant.log files from today and yesterday

Hi,

just noticed, the wpa_supplicant information has been log-rotated. 
Here's the information you probably will like to have:
  var/log/wpa_supplicant.log-20120508
  var/log/wpa_supplicant.log-20120509

Kind regards,
Thorsten

Comment 8 Stanislaw Gruszka 2012-05-10 10:44:12 UTC
This problem is fixed on RHEL6.3, but we need to find a patch for RHEL6.2.

Comment 10 Thorsten Hesemeyer 2012-05-10 11:15:02 UTC
Hi Stanislaw,

agreed, tested RHEL63 starting yesterday - issue did not occur:
  RHEL62 - usually occurs after a some minutes or one/two hours
  RHEL63 - does not occur even when roaming more than 24h

Kind regards,
Thorsten

Comment 11 Stanislaw Gruszka 2012-05-15 10:17:11 UTC
This issue is not a bug on RHEL6.2 release, but rather bug in one of my test patches, claiming to fix 814674 (wireless roaming test on raw RHEL6.2 kernel will trigger a crash).

There was a race condition that mark scanning as pending, but do not perform actual scan and scan pending state was never unmarked. New scan request finished with -EBUSY (Resource temporarily unavailable) error. Hence after some time there was no AP visible except currently associated one, since kernel internally time out old APs information.