CFT: XO-1 wireless scan results loss

[James Cameron]

Call for testing of the XO-1 wireless scan result loss problem.

Purpose: determine if the loss of scan results occurs in the wireless
device, or the operating system kernel.

Method: set up a challenging environment, and perform scans using Open
Firmware only, and then the operating system only.

Ingredients:

1.  several XO-1, with mesh not disabled, with Q2F19,

2.  one more more access points that do not reliably appear in the
    XO-1 network neighbourhood view on reboot of the XO-1s,

3.  a USB drive, with the attached file wsl.zip unzipped on it; you
    might label it "wireless scan logger",

4.  the Linux version of the similar test, see the shell script below,

5.  about two hours of test run time, but more hours yields more
    reliable data,

Recipe:

1.  record the SSID of the access point, and customise the shell
    script accordingly,

2.  record the MAC address of the access point,

3.  make sure the XO-1's individually connect to the access point, and
    use "discard network history" if necessary to ensure it,

4.  turn off all XO-1,

5.  select one XO-1 to run the "wireless scan logger", and leave it
    off for the moment, and turn on each of the other XO-1s, let them
    either connect to the access point, or let them use mesh on
    channel 1, either is fine,

6.  on the XO-1 to use as "wireless scan logger", boot it from the USB
    drive on which you unzipped wsl.zip, and leave the USB drive in
    the XO-1, and leave this XO-1 running for the first half of the
    test,

7.  at the half way point in your testing (e.g. after an hour), press
    any key on the "wireless scan logger" and it will stop, then
    remove the USB drive and copy the new file you will find,

8.  without changing the network configuration, boot the XO-1 into
    Linux, and repeat the test using the shell script, for about the
    same amount of time,

9.  send me the log file from the Open Firmware scan, the log file
    from the Linux scan, the access point SSID, and the access point
    MAC address, and how many of each type of XO were used.

Analysis: by counting the number of times the SSID is missing from the
scan results, and comparing the Open Firmware test against the Linux
test, we can deduce whether the firmware is properly reporting the
scan results to the host.

Optimisation: if your environment is particularly quiet, and you wish
to generate more challenging traffic, and you don't have other people
you might disturb, take the file abn.zip, unzip on it another USB
drive, label it "adhoc beacon node", and boot as many other laptops as
you can with it; XO-1.5, XO-1.75, and XO-4.  Some XO-4 don't run this
tool for long.  The tool will generate beacon traffic on the channel,
and additional probe responses in reply to the active scan by the test
laptop.

--

#!/bin/bash
MA=5C:63:BF:D8:F6:C0
while true; do
    T0=$(date +%s)
    if [[ $(( $T0 % 5 )) != 0 ]]; then
        sleep 0.1
        continue
    fi
    R0=$(sudo iwlist eth0 scan 2>/dev/null | awk "BEGIN{x=0;m=1} /$MA/{x=1;m=0} /Last beacon/{gsub(\"ms\",\"\"); if (x) print \$4} /IE: Unknown/{x=0} END{if(m) print \"missed\"}")
    if [[ "$R0" == "missed" ]]; then
        echo missed
        sleep 3
        continue
    fi
    echo $T0 $R0
    echo $T0 $R0 >> scan.log
    sleep 3
done

-- 
James Cameron
http://quozl.linux.org.au/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: wsl.zip
Type: application/zip
Size: 1108 bytes
Desc: not available
URL: <http://lists.laptop.org/pipermail/devel/attachments/20140210/65b961a1/attachment.zip>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: abn.zip
Type: application/zip
Size: 1102 bytes
Desc: not available
URL: <http://lists.laptop.org/pipermail/devel/attachments/20140210/65b961a1/attachment-0001.zip>