Canadian TV, Computing and Home Theatre Forums banner

Mythtv

76K views 397 replies 31 participants last post by  mwalma 
#1 ·
Mythtv has announced version 0.22 release candidate 1 is now available. Hopefully all goes well and a stable 0.22 will come out shortly.

The release notes are here.
 
#72 ·
Its a bit complicated right now due to some personal time contraints, but I was running 2 pcHDTV PCI ATSC cards with 2 H150 NTSC analogue cards and 2 950Q USB ATSC devices. I've pulled out the 2 pcHDTV cards for now to better assess the 950Qs, but I'm finding that the pcHDTV cards (Generation 5 chipset) lock onto the signal more quickly and require less signal strength than the 950Qs (Generation 6). I would have expected the opposite so I have more testing to do. In the real world this means that ATSC Live TV on the mythfrontend box seems to be hit and miss as a result. This was never a problem with the pcHDTV cards so I'll be putting them back in. OTOH all test recordings made with the 950Qs have been excellent.

So, its not really about the MythTV 0.22 software, its about my own hardware testing. I'm quite pleased with 0.22 actually.
 
#74 ·
The generation of the decoder chipset is only one piece of the puzzle. Tuner sensitivity is another. The higher generation chipsets will handle various types of interference (including multipath) better. For weak signals, the tuner sensitivity becomes more important. Maybe the pcHDTV has a better tuner than the 950Q.

You could always use both the pcHDTV cards and the 950Qs and configure the backend to use the pcHDTV cards for live TV and your weak broadcasts and the 950Q for recording the stronger stations.
 
#75 ·
roger1818, yep, MythTV allows you to prioritize the input devices and that's what I've done in the past. I think you are correct about tuner sensitivity, but I'm in a great location for the Mt. Seymour stations so to me it is very odd that the 950Qs are not performing as well as the pcHDTV cards. :)

mlord, I haven't gone on the pcHDTV forums in quite awhile (product hasn't changed in about 3 years) but I'll take a look in there for NF data and maybe email Jack Kelliher for that info if need be.
 
#78 ·
A few other things I thought of. Once again, I can't compare to 0.21, but some things I did notice with 0.22:

  • The manual record feature only lets you record up to 360 minutes. This might be a file size limitation however.
  • Also with manual record, I would like to be able to specify the end time instead of duration (or at least show me the end time while setting the duration).
  • Once a manual recording is set, it won't let you change (or override on a specific day) the channel, start time or duration (or at least I haven't figured out how to do so).
  • The right and left arrows are used to change values, but when it reaches the maximum/minimum value, the key suddenly changes function to go to the next/previous field. This is a pain when holding the key down to quickly get to the maximum value as if you aren't careful you will start destrying all the other values you have set.
  • It probably wouldn't be used very often, but it would be nice if the OSD would show recording time of day. That way if you want to jump to what was recorded at specific time of day, you can do so without having to remember what time you started to record the program (if I started to record at 1:00pm and I am 1:35 into the program and I want to watch an event that starts at 5:00pm, I need to jump ahead 2:25). This is probably just a matter of adding a code to the OSD XML file.

Despite these limitations, I certainly love Mythtv and am looking forward to getting it to work closer to its potential! One thing I do like about Open Source is if I get a chance, I can fix these issues myself instead of having to rely on someone else to fix them for me.
 
#79 ·
Yeah, the things that are broken the worst (arrow keys) are buried deep in the code. I'll fix them myself, though, eventually. No point in whining over them if I don't intend to do something about it all. ;)

If you want to get your HVR-1600 remote control to work perfectly, then just bring your box + remote (nothing else) over here for an hour. I know all about what needs to be done, and even have a mostly automated setup script to fix it.

I use a PVR-250 remote control on the box here, with pretty much the exact same setup.

Good to hear that manual-record works -- it has *never* worked for me.

Cheers
 
#80 ·
Thanks for the offer to help. I may take you up on your offer after the Olympics are over (things are busy and the Myth box is in almost constant use).

I tired to apply your "fix hauppauge remote" patch last fall, but I couldn't get it to work. I then tried to build the latest cx18 driver and totally killed it and ended up re-installing Mythtv from scratch.

Getting the remote to work is probably the biggest thing from a WAF perspective.
 
#83 ·
I am thinking of installing MythTV to have a look. I have recently installed Linux Mint 8 on a PC ...
You could try it. Just keep in mind the following points:
  • MythTV really, really wants a dedicated PC.
  • MythTV makes Linux look bad. Linux is great, but MythTV is horrid to install, configure, and tweak.
  • An hour or less will get most people up and running. It then takes another year or so of tweaking to achieve a reasonable WAF (Wife Acceptance Factor).

Cheers
 
#82 ·
Roger: is your MytvTV box based on Mythbuntu-9.10?

I ask, because I've now got a development system here with that release, so I suppose I could use that update the fix_hauppauge scripts to work better with 9.10.

But only if it's useful to someone. I don't need the scripts here.

-ml
 
#85 ·
Roger: is your MytvTV box based on Mythbuntu-9.10?
Yes it is.

I ask, because I've now got a development system here with that release, so I suppose I could use that update the fix_hauppauge scripts to work better with 9.10.

But only if it's useful to someone. I don't need the scripts here.
If you have the time, that would be great! It is getting very frustrating not having a remote. I keep finding myself looking at my JP1 remote and imagining what buttons would be assigned to what features. ;) Otherwise we can wait until I get a chance to drop by with my Myth box.
 
#86 · (Edited)
Fresh 0.22 install better than sequentially updated 0.19 -> 0.22

My mythconverg database that I've been using since MythTV 0.19 (which the subsequent 0.20 and 0.21 updates have always upgraded to the latest schema) became corrupted yesterday on the mythbackend machine.

When I restored from a clean nightly mythconverg backup it barfed during the first run of mythfilldatabase, complaining of all sorts of table errors. It was beyond repair using methods recommended by MythTV, and I also used MySQL and phpmyadmin repair tools to no avail. I even went so far as to drop the mythconverg database, create a new one using mc.sql, and then restore from a clean backup, but no luck.

Oh well, stuff happens, but I've swapped so many tuners in and out of that box over the past few years that the latest changes might have pushed the db over the edge. I don't feel like doing a regression analysis to find the exact culprit, but I could probably have restored the existing system if I really tried. Thankfully there wasn't anything priceless on the mythbackend machine. BTW I've always moved or burned the priceless recorded stuff asap, just in case.

So, the good thing is that this gave me the opportunity while the machine was down to convert the OS from Mandriva 2008.1 32-bit to Mandriva 2010 64-bit with it's much more recent Linux kernel, test all the HDs for errors, and do a fresh MythTV 0.22 install from a clean, new database. I also decided to run the storage partitions on EXT4 instead of my old favourite XFS just to evaluate the disk I/O performance.

Performance is much improved in almost every measurable way, so all in all this was an afternoon well spent. :)
 
#90 ·
roger1818 said:
I haven't had a chance to drag a network drop into my living room so I can't use guide data and have to do manual records instead. I hope to get that fixed soon though.
If you have wifi available in the house, then I can loan you a wifi-to-ethernet bridge device, that'll connect things up until you run cat5 someday. It simply plugs into the ethernet port of your MythTV box, and connects to your wifi.

It has a built-in web interface for configuring the wifi parameters / password etc. The MythTV box sees it as a pure ethernet cable connection.

Cheers
 
#92 ·
If you have wifi available in the house, then I can loan you a wifi-to-ethernet bridge device, that'll connect things up until you run cat5 someday.
Thanks! If I don't have it dragged in by the time I see you, I may take you up on your kind offer. My only hesitation is band-aid solutions like this often delay the permanent solution. ;) However, it would let me wait until my parents come to visit next month (an extra pair of hands is always useful for a job like this).
 
#99 ·
Okay, I'll fire up the QuadCore box and build two sets of kernels: one 32-bit and one 64-bit.

Edit: Got the 64-bit one up and running. But you [Roger] will have to bring the box over sometime (next week?) to get the 32-bit kernel installed and the remote working.. I don't want to mess up your box with attempting a "remote control install" of it. :)
 
#101 ·
Peachy. I'm busy for the weekend, though. So how about an open invite for Monday or Tuesday evening, this coming week?

Roger, Greg, anyone else who wants. Just bring your Mythbox + remote control over (no display, no keyboard, no mouse, no antenna, no tower ;)), and we'll fix the remotes and other minor issues.

You guys discuss which evening works best, and let me know on Saturday (email or PM).

Cheers
 
#105 ·
Yep, and XFS's delete speed is amazing! :) I always set up the boot partition under either EXT2 or EXT3 then the others under XFS. This time I opted for EXT4 to see what would happen, but after reading the Phoronix article I'm going to back it all up and restore it onto the file system types above.
 
#107 ·
No worries. The remote that came with your HVR-1600 *does* work, just not with the default setup of Mythbuntu.

Roger1818 also uses 32-bit Mythbuntu, same as you, so we'll work out a nice installable package solution here, and then pass it along to you, and to anyone else who needs it (including another buddy of mine).

The approach I use here, for my own PVR-250 remote as well as anyone else's HVR-1600 remotes, is to simply kill off the massively complex LIRC blasphemy, and use the built-in Linux kernel "keyboard" support for the remote controls. Nice and simple, and easy to customize the buttons, repeat rate, etc..

So long as the kernel has, or can be patched for, support of the specific remote, then this solution is an order of magnitude simpler than the usual LIRC abomination.

The 2.6.32 / 2.6.33 kernels support the HVR-1600 remote by default now. But one still needs the userspace script to enable it.

Cheers
 
#108 ·
MythTV and the HVR-1600: gotchas

Speaking of using the HVR-1600 with MythTV.. the more I use mine, the more bugs I find in MythTV and Linux.

Until recently, I hardly ever used the digital tuner side of the card, since I have four other digital tuners here already. But during the 2010 Winter Olympics, the HVR-1600 digital side was used frequently, to capture HD versions of the hockey games and opening/closing cermonies.

Usually for analog, the PVR-250 card grabs the recordings. But again, during the Olympics, the analog half of the HVR-1600 was also frequently required to handle situations where multiple channels needed to be recorded simultaneously.

All of which led up to the Sunday evening closing ceremonies.. when MythTV picked the HVR-1600 for both the analog and digital recordings of the festivities. This was the first time ever here that MythTV tried to start both an analog and a digital recording at exactly the same time on the same HVR-1600 card.

The kernel cx18 chipset driver crapped out with a single, unhelpful log message...

kernel: cx18-0: Unable to find blank work order form to schedule incoming mailbox command processing

...and neither analog nor digital was recorded successfully. I did arrive home in time to notice this, and scheduled a re-recording from the analog channel for the program rebroadcast later last night.

I'll have to go into the code and find/fix that bug now, I suppose.
The other earlier bugs I know of were these:

  • First analog recording after boot is crap. Solution is to do a dummy read from the card as part of the bootup procedure -- works 100% successfully with that.
  • No sound with some analog recordings. Solution is to configure MythTV Recording Profiles to stick to the driver defaults of 48000 audio sampling rate, and 224kb/sec mpeg Layer-II audio encoding. Still testing, but it has been good since the workaround.
Cheers
 
#145 ·
MythTV picked the HVR-1600 for both the analog and digital recordings of the festivities. This was the first time ever here that MythTV tried to start both an analog and a digital recording at exactly the same time on the same HVR-1600 card.

The kernel cx18 chipset driver crapped out with a single, unhelpful log message...

kernel: cx18-0: Unable to find blank work order form to schedule incoming mailbox command processing

...and neither analog nor digital was recorded successfully.
Did you ever find the cause if this problem? Sorry if I missed the answer. I tend not to use the ATSC tuner much since my box often stutters when playing HD programming (not a huge problem as I don't have an HDTV yet). I am playing with transcoding HD programs to SD, but that means I have to wait before watching the program. Once I upgrade my PC, I hope to use the ATSC tuner more.
 
#109 ·
Interesting. Are you using the latest cx18 driver, or the one that ships with Mythbuntu? I gather at least the first of the "earlier bugs" has been fixed, but as you say there is an easy work around.
 
#112 ·
Interesting. Are you using the latest cx18 driver, or the one that ships with Mythbuntu.
I'm using the cx18 driver that is stock in the 2.6.33 Linux kernel, so it is newer than what Mythbuntu is/will-be shipping with.

But all of the tuner-card drivers are developed out of tree, in their own, closed eco-system. So the truly latest drivers have yet to be pushed out into a standard kernel. They can be obtained from an svn (or git now?) repository, if one wants to go really bleeding edge.

This also means they are much buggier than average for kernel stuff, because they don't get the same scrutiny as mainline kernel stuff does.

I don't know if they've really fixed the "first recording" bug or not, and don't really care now that I've got a good and reliable workaround in my system.

Cheers
 
#114 ·
Getting the HVR-1600 remote control to work in Mythbuntu-9.10

Roger and I finally got together last evening, and enabled the Hauppauge remote control of the HVR-1600, on his Mythbuntu 9.10 system.

Not easy.

The idea here, was to get a working ir-kbd-i2c driver installed. I have a patch to make it work with the 2.6.30, kernel, and it works "out of the box" with 2.6.32 and higher. But Mythbuntu-9.10 uses the in-between 2.6.31 kernel, which cannot be easily fixed.

First, we tried the "simple" solution of installing a 2.6.32 kernel, but the syslog was instantly flooded by some ext4 error messages. Not wanting to risk filesystem corruption, we scrapped 2.6.32, and reverted back to the 2.6.31 Mythbuntu kernel again.

Next attempt was to download the latest snapshot of the entire [tuner] drivers subsystem from the v4l-dvb site, which includes a working ir-kbd-i2c driver with my fixes already in place.

Installing the snapshot drivers is normally simple: unpack the archive, do make and then make install, and reboot. But there were several build errors -- Mythbuntu's fault, not the drivers -- and we had to hand edit the config files to work around those.

Once we got it to build, the make install and reboot worked just fine. We then ran my enable_hauppauge_remote.sh script, and had a working remote control.

To make the fix "persistent", we added a call to that script into the /etc/init/mythtv-backend.conf "upstart" file (not the similar file in /etc/init.d/).

Unfortunately, Mythbuntu-9.10 also uses the new and confusing GRUB2 bootloader, rather than the hitherto Linux standard GRUB loader. In the time available to us, we were unable to figure out how to add a static boot stanza for the current kernel version to GRUB2. Well, we could add one, but it always got put at the end of the list rather than the top (default) position. Really useful, that (not!).

So this means that Roger has to be careful not to apply any automatic Mythbuntu kernel upgrades, because those will then become the default boot kernels, replacing our hacked driver stack. Things would still work, just not the remote control.

In theory, one could just rebuild the replacement driver stack again after any upgrade, but since Mythbuntu has broken the build for it, that's not trivial.

Ugh.

I expect this all to just go away with the upcoming Mythbuntu-10.4 release (April 2010), which is based on the 2.6.32 Linux kernel, with a working ir-kbd-i2c driver included by default. The enable_hauppauge_remote.sh script will still be needed, but that's a simple thing to tack on.

Cheers
 
#115 ·
First of all I would like to thank Mark for his patience with me rescheduling this meeting a couple times and his time and expertise getting this to work! I am not sure if I would have had the time or patience to get it working on my own.

But there were several build errors -- Mythbuntu's fault, not the drivers -- and we had to hand edit the config files to work around those.
To bad we didn't document which config files needed to be edited and which lines need to be disabled to make it easier for others to follow in our footsteps.

Once we got it to build, the make install and reboot worked just fine. We then ran my enable_hauppauge_remote.sh script, and had a working remote control.
One thing though, we weren't able to get the auto-repeat to work well. I may try and play with this some more sometime to see if I can get it working better.

I expect this all to just go away with the upcoming Mythbuntu-10.4 release (April 2010), which is based on the 2.6.32 Linux kernel, with a working ir-kbd-i2c driver included by default. The enable_hauppauge_remote.sh script will still be needed, but that's a simple thing to tack on.
Lets hope my hard drive lasts until then. ;) Once Mythbuntu-10.4 is available, I hope to upgrade my hard drive or maybe even my entire system.
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top