[ezechiel1917]

ah cool, thanks guys!

[ursli]

Just wanted to say thanks for all the hours you guys put in, with the latest activeae/vdpau commits cpu usage for me is lower then ever together with increased stability this is really turning into something rather good.

[ryszardzonk]

I just noticed there is more work done https://github.com/xbmc/xbmc/pull/3037 by FernetMenta on vdpau. From commit I read code is meant to get the most out of ION, but would this help make vdpau work on open source radeon driver or there is still work to be done in there within the driver itself implementing missing features like interop?

[fritsch]

This PR is basically the code that makes the ppa in this thread superior. Fernet is shipping it upstream :-)

Concerning your radeon oss - we have now said multiple times, that the driver, though being a good start, misses corefeatures, which need to be implemented upstream first.

[ryszardzonk]

I new form my personal tries with radeon oss about month ago that there is work in the driver still to be done, but more work on the vdpau and few posts in the forum here and there made me think I missed something ^^. Anyways keep it up

[whoppr]

I messed something up pretty bad in my install and rebuilding. I have been using the AMD cat drivers 12.10. I know there are some issues with these drivers beyond 12.10.

Is there a stable release that folks would recommend beyond this and what benefits are there? I will probably stick with 12.10 but was still curious if there is good luck with any later versions

Thanks

[fritsch]

If you can keep Catalyst 12.10, keep it - everything after it (despite the shipped driver for Ubuntu 12.10 (a specially modified version for Ubuntu 9.000)) is broken or it needs some manual work to get it going halfway ...

Note: The 13.6 solution with 13.8 amdxvba.so files is working for 90% of the users - but it has evil performance problems, when you combine it with boblight. This was a user report on IRC.

[Artax]

This PR is basically the code that makes the ppa in this thread superior. Fernet is shipping it upstream :-)

Concerning your radeon oss - we have now said multiple times, that the driver, though being a good start, misses corefeatures, which need to be implemented upstream first.

So after this merge what will be the differences between the main xbmc and this ppa? In a near future the developement wiil switch to the main xbmc o it will continue here?

[fritsch]

Main differences:
- this ppa has SDL dropped and has proper xrandr support
- this ppa has XVBA for amd gpus

Bot are some 1000 lines of code - so will need some time, nothing too fast - i don't think everything gets in to gotham.

[ezechiel1917]

can anyone check if you have also xbmc crash when you press Stop button while LiveTV is playing and DialogPVRChannelsOSD.xml (OSD Channel list) window is active ?

Code:

11:26:53 T:2674916160   DEBUG: GOT ANNOUNCEMENT, type: 1, from xbmc, message OnStop
11:26:53 T:2674916160   DEBUG: Thread DVDPlayer 2674916160 terminating
11:26:53 T:3003111232  NOTICE: DVDPlayer: finished waiting
11:26:53 T:3003111232   DEBUG: LinuxRendererGL: Cleaning up GL resources
11:26:53 T:3003111232  NOTICE:  (VDPAU) Close
11:26:53 T:3003111232  NOTICE:  (VDPAU) FiniVDPAUOutput
11:26:53 T:2768239424  NOTICE: VDPAU::COutput: vdpau gl interop finished
11:26:53 T:2768239424  NOTICE: COutput::OnExit: Output Thread terminated
11:26:53 T:2768239424   DEBUG: Thread Vdpau Output Thread 2768239424 terminating
11:26:53 T:3003111232   DEBUG: CVDPAU::FiniVDPAUOutput destroying 7 video surfaces
11:26:53 T:3003111232   DEBUG: CGUIWindowManager::PreviousWindow: Deactivate
11:26:53 T:3003111232   DEBUG: ------ Window Deinit (DialogPVRChannelsOSD.xml) ------

Also getting xbmc crash when waking up from suspend when "DEBUG: CPVRTimers - Update - updating timers" timers kicks in.
If it is suspended for a while it's immediate crash when system wakes up. If I do suspend and resume fast after that. It wakes up fine but crashes when first occurance of "DEBUG: CPVRTimers - Update - updating timers" happens. In meantime I can play Live TV just fine.

What's the proper place to report these two bugs?

Thanks Ez.

[fritsch]

You are at the right place. But the debug log you posted - is (you might guess it) so minimal that one cannot see anything. There is also nothing wrong in the log.

What do you mean with "crash" - segfault or "hang"?

Proper Debug Log please.

[ezechiel1917]

Sorry for that... I don't have tracing enabled atm, had some hdd space issues with it (will look at it later), so there is nothing more in the log when this happens. xbmc crashes (segfaults I guess because it is restarted immediately) and "DEBUG: ------ Window Deinit (DialogPVRChannelsOSD.xml) ------" is last line there, same goes for resume for suspend issue, last line "DEBUG: CPVRTimers - Update - updating timers".

[fritsch]

Ha, cool - yes, I could reproduce. Thanks for reporting.

[fritsch]

That fixes it:

Code:

diff --git a/xbmc/pvr/dialogs/GUIDialogPVRChannelsOSD.cpp b/xbmc/pvr/dialogs/GUIDialogPVRChannelsOSD.cpp
index 3183394..12eaded 100644
--- a/xbmc/pvr/dialogs/GUIDialogPVRChannelsOSD.cpp
+++ b/xbmc/pvr/dialogs/GUIDialogPVRChannelsOSD.cpp
@@ -148,8 +148,10 @@ bool CGUIDialogPVRChannelsOSD::OnAction(const CAction &action)
CPVRChannelGroupPtr CGUIDialogPVRChannelsOSD::GetPlayingGroup()
{
   CPVRChannelPtr channel;
-  g_PVRManager.GetCurrentChannel(channel);
-  return g_PVRManager.GetPlayingGroup(channel->IsRadio());
+  if(g_PVRManager.GetCurrentChannel(channel))
+    return g_PVRManager.GetPlayingGroup(channel->IsRadio());
+  else
+    return CPVRChannelGroupPtr();
}

void CGUIDialogPVRChannelsOSD::Update()
@@ -190,7 +192,8 @@ void CGUIDialogPVRChannelsOSD::SaveControlStates()
   CGUIDialog::SaveControlStates();

   CPVRChannelGroupPtr group = GetPlayingGroup();
-  SaveSelectedItem(group->GroupID());
+  if(group)
+    SaveSelectedItem(group->GroupID());
}

void CGUIDialogPVRChannelsOSD::RestoreControlStates()

I have to talk with opdenkamp the PVR subtree maintainer. I don't know the workflow within PVR, so - could only make it working. It started with GetCurrentChannel not being checked of being empty and afterwards a null pointer dereference.

Watch PR here: https://github.com/xbmc/xbmc/pull/3118 (merged, will come after next rebase)

[ezechiel1917]

thanks!