I posted this question to other newsgroups before, but nobody answered. It
seems the kernel hackers are listening here. So, please apologize for my
networking quesion here. Our newsfeed is not very reliable, so if you know
an answer, please also mail me directly.
Several users have reported that ncpfs connections are hanging from time to
time. I have not been able to reproduce this, although I tried very
hard. So I could not fix it yet. Sorry that I can not say more about the
problem, I simply was not able to find out more. Maybe someone out here can
help us?
Under certain circumstances I could not really find yet, ipx_sendmsg
(called in fs/ncpfs/sock.c) hangs forever when it is called in blocking
mode. A hint might be that it is likely to happen when the network load is
really high. On one machine this is reproducable when you call ipxdump
after ncpfs has been mounted.
Is this expected behaviour, or can it be called a bug? If it is expected
behavoiur, does anybody know how I can make sure the packet is eventually
sent out to the wire?
BTW, the Caldera NetWare client does not show this problem. The main
difference between Caldera and ncpfs I know of is that Caldera calls the
ipx socket routines from user space, whereas ncpfs calls them directly from
kernel space. But to be honest, I do not know how that can cause any
problem.
Thanks for your time,
Volker
--
Volker Lendecke <lendecke@namu01.Num.Math.Uni-Goettingen.de>
For my PGP public key, finger my email address.