[mythtv-users] Hauppauge HD-PVR trials and tribulations.

[georgia_tech_swagger]

I figured I would post this to the list in the hopes it would help somebody
else as much as me, as most of the similar results I found trying to solve
this problem were old threads on this list.

My HD-PVR was running just fine on my old X58 Core i7 920 system.  I just
upgraded to X99 on a Core i7 5820k.   An my HD-PVR went nuts on the new
machine.

First ... some migration tips:
- Don't forget to make sure your Storage Directory has proper permissions
for user and/or group (depending on distro) mythtv
- Don't forget to include your change channel script if you're controlling
a converter box with a blaster like I am.  If the script isn't there, there
is no error what-so-ever... just silent failure.  That one was a PITA to
figure out.

So ... it would work but hard lock up the HD-PVR, typically in less than 5
minutes, but sometimes taking 15-20 minutes.  I tried the usual myriad of
Google suggestions with mixed success:
- Overheating?  No.  I keep it fairly chilly in the winter.
- Crap capacitors in power brick?  No.  Was working just fine previously.
- Try using irqbalance?  This actually helped double the time to lock but
did not fix it.
- Try using a USB2 instead of USB3 port?  This actually helped double the
time to lock but did not fix it.
- Try disabling USB3 on the motherboard?  I haven't done this.  And REALLY
don't want to unless it is the full on last resort.
- Try upgrading firmware of HD-PVR?   I havne't done this.  I don't have
immediate access to a Windows box.  I believe I'm using 0x1e
- Make sure you're not using an old kernel?   I tried 3.16.5-gentoo first.
Horror show.  I am now using 3.17.6 ... which seems to greatly improve
things.  In my first test I got up to about 40 minutes before lock up.  I'm
in the second test now.

The logging from mythfrontend, mythbackend, and modprobe hdpvr
hdpvr_debug=7 aren't particularly helpful.   The HDPVR in debug would
return:

config call request for value 0x800 returned -110

Right before that happened you'd get a steady stream of calling for buffer
1 ... but immediately before that error you'd get call for buffer 1
immediately followed by call for buffer 2.  And that is, as far as I can
see in the blur of those calls going by, the only time that happens...
right before the error.

In mythbackend you would get:

"Poll giving up 2"
"Device error detected "

I welcome all suggestions/advice/debug tips.  It seems most likely to be a
kernel issue in the driver and/or xhci.   My old X58 system was on kernel
3.10 ... but I cannot downgrade that far as I need at least kernel
3.15/3.16 for critical patches for the soundcard on my Asus X99 Pro
motherboard (ALC1150 Intel HD).
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mythtv.org/pipermail/mythtv-users/attachments/20141220/f8370fe7/attachment.html>