Mythtv incompatibility with gcc 431

[KungFuJesus]

So I completely rebuilt a working mythtv system after a reformat. I got through the build process just as it was before, but this time accepting ~x86. In most circumstances, this doesn't make too much of a difference, seeing as most all of my multimedia packages were unmasked to use testing anyway. The ~x86 keyword forces me to use nvidia's later drivers (the newer kernel 2.6.25 dictates it). So, building this, I find out mythtv just crashes when it uses xvmc, and when using xv, it's not as good as it could be. I carefully follow the mythtv wiki only to find out the crashing error is because I didn't add an addition NvAGP "1" option to my xorg.conf. It turns out what this does is force the video card to use Nvidia's AGPGART driver. Well, ok, so I can finally watch mythtv. I start it up only to find that there are hundreds of audio buffer underruns filling the mythfrontend log. I try mplayer with XvMC, it works fine. I try video without XvMC on mythtv's internal player: works even worse than before. I enable Side Band Addressing and Fast Writes (keep in mind this is a 7600 GT, a video card that is more than capable of HD quality video). This makes things a little better, but still unbearable. I'm not going down a kernel version as well as an nvidia driver version in hopes that this fixes things. My hypothesis: Nvidia's drivers broke something between 167 and 169-up. If this works, it only makes me wonder if the mythtv devs may have been better off with an open source implementation of nvidia's drivers written by nvidia much like how unichrome's XvMC works...

[yabbadabbadont]

Do you have an actual support question, or are you just ranting?

[ali3nx]

[Akkara]

[KungFuJesus]

I will scour my kernel. A simple downgrade of kernel and nvidia drivers doesn't seem to have done it. The other subtle difference I find is that I was running Xorg 7.2 before and not 7.3, so I'm rolling all of that back a version to see if this helps things out. I'm using alsa, although it seems to be complaining about the mixer device (not that it matters, I'm using the digital passthrough anyway, it may as well not even be functioning as a soundcard).

I have preempt voluntary, but with real-time priority it shouldn't really matter, should it?

[KungFuJesus]

Hmm, the issue seems to be with mythfrontend's internal player. Mplayer works perfectly. With playback turned on verbose for the logging, I get the following errors:

[KungFuJesus]

I GOT IT! GCC 4.3.1 produces unexpected mythtv binaries. It literally borked my performance to something unusable. Hmmm, I'll work at a bug report after more investigation.

[tld]

[KungFuJesus]

But this is why they're marked testing, so that we can test them and report back the bugs .

[yabbadabbadont]

[i92guboj]

While testing is acceptable, and meant for reporting bugs and solving issues, I have not seen any of that on this thread. It just started as a rant about how using the unstable branch gives you erm... well, a unstable system.

However, since the thread has lately evolved to something constructive, I will leave it where it is for now. Maybe someone with a similar problem can find it useful.

Please, consider changing the title to something that actually describes the issue, because as it is right now it's not only unrelated, but also meaningless. The change the title of the thread edit the first post and change it's title.

Thanks.

[KungFuJesus]

[i92guboj]

[KungFuJesus]

Only thing I changed was the title :-p. I file my bug reports when I have the time, . I must say that I did file that bug report before you posted your post, though