[pulseaudio-discuss] Module-tunnel unloading itself - protocol error, kicking client.

Colin Guthrie gmane at colin.guthr.ie
Mon Dec 3 13:51:08 PST 2007

Colin Guthrie wrote:
> Colin Guthrie wrote:
>> Colin Guthrie wrote:
>>> Client is x86_64 and client is i586, but I've played music via a tunnel
>>> all day today with the same setup at work. The server at home is an EPIA
>>> in an NFS root fs (read-only). Any thoughts/musings welcome :)
>> For the benefit of others:
>> <coling> mezcalero, and anyone else. Any idea how to debug the issue
>> with tunnel's dying due to protocol errors? I'm happy to dig into it but
>> woudl appreciate some pointers to try and narrow it down (reported on ML)
> snip.
> Perhaps too premature a fix?
> I'll just past what I posted on IRC but was too late at night to catch
> anyone ;)
> <coling> mezcalero, neurocyte did you guys get the tunnel working with
> just that simple patch?
> <coling> I seem to get a nice new error now...
> <coling> E: pdispatch.c: Recieved unsupported command 77
> <coling> E: module-tunnel.c: Invalid packet
> <coling> I also get an error with command 76
> <coling> Any ideas?
> Have you guys managed to get PA 0.9.8 + patch working? Perhaps we need
> to ignore unknown commands in the tunnel module to get this hack working?

There is an incompatibility with auto suspending. If the server to which
the tunnel is created suspends it's audio device, it will kill the tunnel.

I've "fixed" this by ignoring suspension (the assumption being that it
will automatically open it again).


This patch should be applied to pa 0.9.8 packages.



More information about the pulseaudio-discuss mailing list