[mythtv] DEC2000-t scan hangs - more info

[Joe]

Hi - I seem to have come up with exactly the same workaround myself last
night!

I have been running a hacked version with both a PCI Nova-T and a
DEC2000-t on USB, and it seems to work fine for recording and live-TV
etc, though I haven't fixed the channel scan for Dec either - I just
used the Nova-t to fill my database.

Having looked at tzap and some of the linuxtv api docs, I would guess
that this is a valid fix - they don't seem to set up the demux using
DMX_PES_OTHER. From what I can understand, the PCI Nova-T driver does
handle it 'correctly' if the PID is sensible, but I would bet that the
DEC2000-t driver isn't alone in not supporting it - there seem to be
quite a few posts about DVB-T cards that still work on tzap/dvbstream
but not on Myth 0.17...

Hope that helps,
		   Joe.

-----Original Message-----
From: mythtv-dev-bounces at mythtv.org
[mailto:mythtv-dev-bounces at mythtv.org] On Behalf Of Jim
Sent: 22 February 2005 22:28
To: mythtv-dev at mythtv.org
Subject: Re: [mythtv] DEC2000-t scan hangs - more info


> Hi - this mail refers to an original message posted by Frank Muenchow 
> regarding problems with a DEC2000-t DVB-t device.  I have what appears

> to be the same problem, but this time in the UK.  Also, my PC has a 
> second tuner - a Hauppage WinTV Nova-T - which is working perfectly, 
> so I suspect a problem specific to the DEC2000-t here.

I have the same issue - 

The scan code doesn't appear to set up the correct entries in 
dtv_multiplex - the set of frequencies (apart from the first two) are 
wrong - I don't know why yet. I had to build channel and dtv_multiplex 
tables manually from a 0.16 dvb_channels/dvb_pids. Having set these up 
verbose logging shows the card gets opened - all the EIT data and the 
correct pid values etc are logged but I still got "No data from card in
1 
second"

I finally get the Dec 2000 to work by hacking the
DVBRecorder::OpenFilters 
method. 

When _software_filter_option is false this function calls ioctl on the 
device with DMX_SET_PES_FILTER for each PID - unfortunately it always 
calls the ioctl with a pes_type of DMX_PES_OTHER - which won't set up
any 
data feeds from the DEC 2000. 

The passed ES_type is no longer compatible with the parameter type of 
dmx_pes_type so a switch on the passed ES_type is needed to set the 
correct type in the parameters

If this is added then everything works ok (recording, live tv etc)

I'm reporting this now because I don't understand the intention behind
the 
0.17 design and someone else might. I don't want to suggest a patch
until 
I've got some confidence it won't break anything.