Hi All.
Following last weeks discussion I added a bidirectional test to l2test - 
this should make it possible for me to set up symmetric multislot 
connections, right?

l2test is running out of short commandline switches, so I chose -x, -y in 
lack of anything more intuitive. Most of the code is duplicated from 
elsewhere in l2test. It's missing a commandline switch to select the 
amount of packets to receive before computing the throughput. The 
sendbuffer is the same as for the "-s" sender and the receive buffer is 
set by the "-I" option. Since I've gotten this wrong more than once, 
comments are wellcome (wether this does what I think it does or not =)..

As I'm working at home today I havn't been able to try the same devices as 
I did last week (will do tomorow), so I tried with a BrainBoxes PCMCIA 
card and a Bluefrom USB unit.

First test: Same as last week (the Bluefrog box doesn't work with 
multislot packets for some reason).

Sender (BrainBoxes):
./hciconfig hci0 ptype DM1;./l2test -b 2000 -s 00:00:00:00:20:17
...

Receiver (Bluefrog):
./hciconfig hci0 ptype DM1;./l2test -b10000 -r -I 2000
l2test[1233]: Waiting for connection on psm 10 ...
l2test[1236]: Connect from 50:C2:C0:03:C2:8F [imtu 2000, omtu 672, 
flush_to 65535]
l2test[1236]: Receiving ...
l2test[1236]: 10000 bytes in 1.24 sec, 7.87 kB/s
l2test[1236]: 10000 bytes in 0.92 sec, 10.59 kB/s
l2test[1236]: 10000 bytes in 0.91 sec, 10.74 kB/s
l2test[1236]: 10000 bytes in 0.90 sec, 10.89 kB/s
l2test[1236]: 10000 bytes in 0.87 sec, 11.25 kB/s
...

With my new patch:
One host (Bluefrog):
./hciconfig hci0 ptype DM1;./l2test -b2000 -I 2000 -x
l2test[1302]: Waiting for connection on psm 10 ...
l2test[1302]: Connect from 50:C2:C0:03:C2:8F [imtu 2000, omtu 2000, 
flush_to 65535]
l2test[1302]: Received 12000 bytes in 2.86 sec, 4.10 kB/s
l2test[1302]: Received 12000 bytes in 2.80 sec, 4.19 kB/s
l2test[1302]: Received 12000 bytes in 2.47 sec, 4.74 kB/s
l2test[1302]: Received 12000 bytes in 1.83 sec, 6.39 kB/s
...

On the other host (BrainBoxes):
./hciconfig hci0 ptype DM1;./l2test -b2000 -I 2000 -y 00:00:00:00:20:17
l2test[1592]: Connected [imtu 2000, omtu 2000, flush_to 65535]
l2test[1592]: Received 12000 bytes in 0.43 sec, 27.05 kB/s
l2test[1592]: Received 12000 bytes in 0.91 sec, 12.94 kB/s
l2test[1592]: Received 12000 bytes in 0.37 sec, 32.04 kB/s
l2test[1592]: Received 12000 bytes in 0.20 sec, 59.82 kB/s
l2test[1592]: Received 12000 bytes in 0.43 sec, 27.41 kB/s
l2test[1592]: Received 12000 bytes in 0.43 sec, 27.09 kB/s
l2test[1592]: seq missmatch: 41 -> 42
...


So after all this trouble I still don't have clue what's going on!? If I 
reverse the "connector" (-y) and the "listener" (-x) both get about 
10kB/s. On several ocations the Bluefrog device has been shown to be ill 
in other ways, could this be it again or is there some blunder in my 
biderectional test?


-- 
Regards Martin Leopold.
Dept. of Computer Science, University of Copenhagen