[mythtv-commits] Re: Ticket #343: 'Signal 100% (l ) no lock' and no video with DVB-S and DVB-T

[MythTV]

#343: 'Signal 100% (l ) no lock' and no video with DVB-S and DVB-T
-----------------------------------+----------------------------------------
 Reporter:  daniel.danner at gmx.net  |        Owner:  danielk
     Type:  defect                 |       Status:  new    
 Priority:  major                  |    Milestone:  0.19   
Component:  mythtv                 |      Version:  head   
 Severity:  high                   |   Resolution:         
-----------------------------------+----------------------------------------
Comment (by adrian.wilkins at gmail.com):

 I'm (now) using a rather recent -mm kernel, but my kernel upgrades of late
 have been in hope of fixing this issue ; I've previously had all three
 tuners working pretty reliably (oh why did I change Myth versions ;-)  ).

 I think your 2.6.9 kernel must have been patched, because I'm fairly sure
 that I had to upgrade my kernel from one newer than that in order to
 support my Nova-t USB2 (we have pretty much the same tuners, but I have a
 DEC2000-t as well).

 If tzap reads it's output regarding signal parameters from the drivers,
 than this has always looked like a driver issue to me ;

 {{{
 using '/dev/dvb/adapter1/frontend0' and '/dev/dvb/adapter1/demux0'
 reading channels from file '/root/.tzap/channels.conf'
 tuning to 834166670 Hz
 video pid 0x0b86, audio pid 0x0b23
 status 1b | signal 612f | snr 0000 | ber 001fffff | unc 0000ffff |
 FE_HAS_LOCK
 status 1f | signal 6135 | snr 0000 | ber 00000000 | unc 00000000 |
 FE_HAS_LOCK
 status 1f | signal 0000 | snr ffff | ber 00000000 | unc 00000000 |
 FE_HAS_LOCK
 status 1f | signal 0000 | snr ffff | ber 00000000 | unc 00000000 |
 FE_HAS_LOCK
 status 1f | signal 0000 | snr ffff | ber 00000000 | unc 00000000 |
 FE_HAS_LOCK
 status 1f | signal 0000 | snr ffff | ber 00000000 | unc 00000000 |
 FE_HAS_LOCK
 }}}

 The channel chosen should have forced tuning to a different multiplex. If
 tzap is reading parameters from the driver, and the signal monitor is
 reading the same things, then while the ber (bit error rate) and unc
 (incorrected blocks) values are good, along with the snr (signal-noise
 ratio), the signal is reported as zero, which I'd regard as a bad sign if
 it weren't for the card being capable of recording perfectly at the
 command line in this condition. In addition, the first two lines of output
 usually represent the tuning pause ; on both my other cards, these two
 lines lack the FE_HAS_LOCK flag. The snr value would appear to agree with
 this ; it looks like the driver is misreporting the signal paramters.

 You can see tzap output from all three of my cards in the archive attached
 to this thread. Only the dvb0, or the PCI card, has an output that I would
 consider consistent and sensible (and even that is not 100% right, because
 the signal value appears fixed even when tuning). dvb1 reports lock to
 tzap, but the log says that Myth does not see it (same bat-time, same bat-
 channel). And for dvb2, Myth reports that it cannot measure several signal
 parameters ; I'm not sure what tzap does if a driver genuinely does not
 report these values, but the output appears to show it as present. dvb2
 then fails over PAT and PMT. This is also what dvb1 does when disabling
 signal monitoring by patching SignalMontitor::IsSupported() to return
 false.

 So point one is that the signal monitoring code is probably preventing
 dvb1 from working because it's falsely reporting bad values (either
 through bugs in driver or sigmon code). Alas, patching sigmon off does not
 fix dvb1 ; it then starts reporting PMT failures.

 Perhaps the output from these USB devices isn't supporting PAT and PMT?
 Isn't it the case that it isn't even a full TS? I mean, for USB 1.1, a
 full TS is never going to cross the bus in a timely manner ; it's a
 struggle sometimes to get a single PS across it (esp. on high-bandwidth
 shows). Do the drivers even support TS, and if they do, are they retro-
 constructing a "poor mans" TS with no PAT/PMT?

 To test this assumption later, I'll patch SignalMonitoring off again, and
 then set dvb1 up with TS recording off - TS ''on'' has become the default
 (from the previous default of PS) and the option to configure it has been
 removed from the GUI. Anyone in the know could tell us whether this
 setting still actually has an effect :-)

-- 
Ticket URL: <http://svn.mythtv.org/trac/ticket/343>
MythTV <http://www.mythtv.org/>
MythTV