This bug reports applies only on the new zd1201 driver which
was posted on the forum, not to the modified wlan-ng driver.
The connection can be frozen by packets not correctly sent by
the zd1201 driver. These packets freeze the other end (in ad-hoc
mode), until another packet is sent to this end. The connection
can therefore be kept alive by permanently pinging the other end
from the machine executing the zd1201 driver.
It appears that DNS replies quite often trigger this bug.
My network architecture is as follow:
ISP gateway laptop/client
x.x.x.x 192.168.20.1 192.168.20.2
DNS bind bind
My both bind nameserver only caches DNS replies. They both
forward my ISP DNS, and can query root DNS too if necessary.
That is, the client bind daemon doesn't forward the gateway bind.
Attached are the tcpdump from both the gateway and the client
which will be frozen. The connection was lost at 10h15. The test
sequence is the following:
from client :
1. ping 192.168.20.1 permanently
2. DNS query (10h15).
3. Connection lost.
4. Clients tries to ARP after time-out.
5. ping 192.168.20.2 to restore the connection (10h17)
6. Both are pinging each other
7. Stopping gateway pings
8. Stopping client pings
end of sessions.
Both computers had the right time.
Log in to post a comment.