[mythtv] 0.21 Firewire bus reset every 60 seconds?

Michael Cunningham mcunning at yahoo.com
Fri Oct 3 00:40:23 UTC 2008

I have this happen ALL the time, it will work great for days and then I'll get recordings that drop about 1 second every 60 seconds.  Sometimes it will go into an endless loop of bus resets for about 20 min and the backend will die.  I cringe when my wife gives me "the look" when we begin to watch one of her favorite shows and it starts to "glitch."

I've tried lots of things to try to get rid of it, nothing seems to have made much of a difference.

- upgrade to more recent kernel - custom build
- swap cable boxes with the cable co (RCN- 6200)
- different cable
- upgraded libraw1394 headers - recompile mythtv
- update raw1394 and other 1394 kernel modules

Right now I have a cron job that runs every hour and checks to see if there is an active recording.  If there is none, it runs firewire_teser -B -n 1.  I also have nagios run an event handler when the backend dies and first runs firewire_tester.  This seems to have helped but really is just a fancy work around.

I noticed that the raw1394 support myth uses seems to be deprecated, and all development seems to be in the new branch (juju?) perhaps migrating to that might be the best way?  Its beyond my programming skills as of today.

I know that thanks to 5c there isn't a whole lot of firewire users out there, but RCN has yet to put it on most of their channels and it is a great way to get content, and its something MCE hasn't been doing as well.  I'm hoping that someone can find a fix for this.


----- Original Message ----
From: Eric Smith <erictsmith at gmail.com>
To: Development of mythtv <mythtv-dev at mythtv.org>
Sent: Wednesday, October 1, 2008 1:37:43 PM
Subject: Re: [mythtv] 0.21 Firewire bus reset every 60 seconds?

On Wed, Oct 1, 2008 at 10:22 AM, Derek Atkins <warlord at mit.edu> wrote:

"Eric Smith" <erictsmith at gmail.com> writes:

> On Mon, Jun 16, 2008 at 10:49 AM, Derek Atkins <warlord at mit.edu> wrote:
>     Hi,
>     Derek Atkins <warlord at MIT.EDU> writes:
>     > Here's an example of the backend log between the 60-second "okay
>     period":
>     >
>     > 2008-06-11 08:43:37.498 LFireDev(000E5CFFFE5AF9A5): SignalReset(86->87)
>     > 2008-06-11 08:43:37.553 LFireDev(000E5CFFFE5AF9A5): SignalReset(86->87):
>     Updating device list -- begin
>     > 2008-06-11 08:43:37.815 LFireDev(000E5CFFFE5AF9A5): SignalReset(86->87):
>     Updating device list -- end
>     > 2008-06-11 08:43:37.917 LFireDev(000E5CFFFE5AF9A5), Warning: No Input in
>     50 msec...
>     > 2008-06-11 08:44:37.501 LFireDev(000E5CFFFE5AF9A5), Warning: No Input in
>     100 msec...
>     > ...
>     After updating libraw1394 from 1.2.1 to 1.3.0 this kind of error seems
>     to have been reduced..  But it also seems to be a problem if the
>     firewire broadcast bus isn't completely synched.  For example, the
>     myth box was rebooted yesterday and then the first recording had
>     this issue.  But then I reset the bus manually using firewire_tester
>     (which, by the way, took two or three tries before it synced!)..
>     and then my recording this morning was fine.
>     So something weird is still going on, but at least the libraw upgrade
>     seems to have helped.
>     Any other suggestions?  My next step is to setup a script to run
>     firewire_tester on boot-up to make sure the bus is stable before
>     mythbackend runs.
>     -derek
> I also see this issue, somewhat infrequently, but enough to be a problem.
> Does anyone have any suggestions on things I could check to resolve it?

I converted over to an external channel change script that runs
firewire tester on every channel change to force a bus synchronization.
This might reset the bus, so it might not work right if you have
multiple firewire devices.   The script is based on the script
from http://www.mythtv.org/wiki/index.php/User:Steveadeff#6200changer.sh
but I had to make a few changes to get it working right for me.

Since I've changed to this script I haven't lost a single recording,
and the number of times I get the sequencial bus resets from myth
has been reduced dramatically (although not completely eliminated --
I've still seen it happen once or twice).

Another side effect of this script is that LiveTV times out when
using this tuner.  Apparently the script takes "too long" and
LiveTV gives up too quickly.  I haven't looked into extending
the timeout.


      Derek Atkins, SB '93 MIT EE, SM '95 MIT Media Laboratory
      Member, MIT Student Information Processing Board  (SIPB)
      URL: http://web.mit.edu/warlord/    PP-ASEL-IA     N1NWH
      warlord at MIT.EDU                        PGP key available

I can certainly give that a try and see what happens.  I only have the one box on that firewire bus so that wouldn't be a problem.
However, that's a bit of a band-aid isn't it?  I'd be interested to understand the root cause of the issue.  It certainly appears that the data stream is coming through fine at the point myth issues the reset.  So where is the missing link that is causing myth to issue the reset?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mythtv.org/pipermail/mythtv-dev/attachments/20081002/a67848cc/attachment.htm 

More information about the mythtv-dev mailing list