[Asterisk-Users] DTMF crashes chan_capi

[Jamie Neil]

Quoting Myself :(
> Hi,
>
> I'm having a problem with DTMF tones from my SIP client
> apparently crashing
> the chan_capi driver. However I'm not sure whether this is a bug or
> misconfiguration on my part: if I set "softdtmf=1" in
> /etc/asterisk/capi.conf the problem goes away. Does the AVM B1 not support
> DTMF detection?
>
> The set up I have is using latest CVS (3 days old) running RH8 on a 933MHz
> P3. SIP client is SJPhone (G711) on Windows connected to * via a 100Mbit
> switched LAN. PSTN connection is using latest chan_capi on an AVM
> B1 PCI v4
> card (a steal at £1.20 on ebay :)).
>
> I can call the * box using the SIP client and interact with the voicemail
> app with no problems using in-band DTMF. I can also call in from the PSTN
> through the capi interface and interact with the IVR menu with no
> problems.
> Finally I can bridge the CAPI and SIP channels and hear DTMF
> digits entered
> on the PSTN phone with no problems (they are also detected and
> displayed on
> the console).
>
> However when the CAPI and SIP channels are bridged, entering more than a
> couple of DTMF digits into the _SIP_ client appears to crash the channel:
> neither party gets disconnected, but there is no longer any audio
> in either
> direction and new calls (inbound or outbound) trying to use the
> CAPI channel
> fail. Once locked if I enter "capi info" in the * console it
> return nothing
> and trying to autocomplete capi commands e.g. "capi [TAB]" just locks the
> console up. Entering capiinfo and lsmod at the command prompt suggests the
> driver is ok. The only way of getting it working again is to restart *.
>
> When I switch to softdtmf, everything seems to work fine, but I
> noticed that
> even though DTMF signalling works fine on the IVR menu, once the call is
> bridged DTMF digits entered on the PSTN phone are not displayed on the
> console like before.

Further to my earlier post...

I've replicated this problem using a passive Fritz PCI card, so my bargain
B1 card appears not to be the problem.

I can only replicate the lock ups when bridging a call directly from
chan_capi to chan_sip (or vica versa) and I use in-band DTMF detection in
the SIP channel. I have tried various other combinations, including SIP>SIP,
CAPI>CAPI, CAPI>MEETME>CAPI and CAPI>MEETME>SIP, and they work without a
problem. The SIP client I am using is SJPhone (which does only in-band
DTMF), and if I switch off the in-band detection in sip.conf it works fine,
but of course I can no longer signal to * from the SIP phone. I can't test
whether out of band signalling is ok as SJPhone doesn't support this.

I find it hard to believe that this is a capi problem - I would have
expected see some other capi related problems aside from this single (easily
repeatable) error.

Jamie Neil
Versado I.T. Services Ltd.