[mythtv-users] Upgrade from 0.26 to 0.27 broke firewire

[Peter Bennett (cats22)]

On 07/01/2014 11:59 AM, John P Poet wrote:
> On Mon, Jun 30, 2014 at 4:22 PM, Peter Bennett <cats22 at comcast.net
> <mailto:cats22 at comcast.net>> wrote:
>
>     I have struggled with the firewire issue - megabytes of "no input
>     in xxxx msec" in the log and no recording. It seems to be random
>     as to when it happens. I had it consistently for 6 months then it
>     stopped. That was with mythtv 0.24. It stopped for a year or so
>     then started again on mythtv 0.26. At that time somebody on this
>     forum recommended changing to a firewire card with Intel chipset
>     instead of VIA chipset. Evidently there was a bug in the firewire
>     Linux driver for that chipset. Since I changed to a card with
>     Intel it has not happened.
>     My fix for when this happens is a script that restarts the
>     backend. If you kill and restart the backend, it picks up the
>     recording again and does not have that error again in that
>     recording. (sudo restart mythtv-backend in Ubuntu).
>     My script continuously monitors the log file and kills the backend
>     when those errors happen for a couple of seconds. Very little
>     recording is lost but the recording is split into two files.
>     Look here for my scripts
>     https://code.google.com/p/mythtv-scripts/
>     The monitor script is here
>     https://code.google.com/p/mythtv-scripts/source/browse/trunk/install/opt/mythtv/bin/monitor.sh
>      
>     Peter
>
>
> Hi Peter,
>
> I don't believe any of the MythTV devs have firewire anymore.  Most
> people that did have switched to cable card.
>
> That being said, if you post logs from mythbackend with "-v
> channel,record" I will look and see if I can spot any obvious
> problems.  One of my TODO list items is to rewrite the firewire code
> to conform to the multirec method used by the other input types --
> however I am not looking forward to it, since I will have no way to
> test it...
>
>
> John 
>
Hi John
I think the best person to do that is the original poster here, DaveD.
My firewire has been working perfectly since I switched to the Intel
chipset Firewire card.

Look at this ticket
https://code.mythtv.org/trac/ticket/10277

I actually created a patch for this in Mythtv 0.24, that solved the
problem. The patch failed in 0.25 and the problem had gone away for me
by then. The patch was a kludge work around - i.e. closed and opened the
firewire port if the error occurred. That action recovered and recording
could continue.
There is a comment in the patch from andy that suggests that the issue
may be related to the VIA chipset.
When I was having the issue I did look at verbose logs and could not
find any cause. It seemed that after recording for a few seconds it
would stop receiving data. One thing - the bus reset that is in the code
and gets called every few seconds when this error is happening - does
not fix the problem, but seems it may make it worse. I tried with that
option both on and off and that did not help.

The general view from most mythtv developers seems to be that firewire
is dying out. I think the cable companies now have a waiver that lets
then give a DRM protected ethernet stream instead of firewire.

Let me know if you need some help testing firewire changes.

Instead of putting more time into firewire, is anybody investigating
HDMI capture? There are HDMI capture cards available now and I have seen
a couple with Linux drivers. They seem to be expensive. I know that
would not work if the HDMI was encrypted, however Comcast has the small
"adaptors" with HDMI output and I wonder if they use encrypted output.
How does one check if the HDMI stream is encrypted?

Peter

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.mythtv.org/pipermail/mythtv-users/attachments/20140701/54845c47/attachment.html>