- Thread starter
- #31
After another evening of debugging I'm now completely at a loss for a solution
I've deployed MP 1.1.0 Alpha on my development laptop with Vista Ultimate. The fix I suggested in this topic solves the problem of the WinTV CI device not being succesfully inserted into the graph, however on all my systems (XP, Vista and Win7 now ) the issue remained that it couldn't finish the graph because of the MPEG2 demultiplexer not being connected. I debugged all the graphs being produced, switched some of the pin logic around and that worked fine.
So I got it to the point where my test code scans and detects non-FTA and FTA channels without crashes or errors, which is great. HOWEVER.....
My test code scans 754MHz only band because I know channels are there in my area, while TVServer obviously scans all possible frequencies. If I deploy my custom tvlibrary.dll into the live TVServer I still get a BSOD after 'some time of scanning'. Yesterday after 12 scans it BSOD'd in hwc11bda.sys, which is the CI-driver. After I added some delays in the code between scans, it made some 20 scans before crashing in the same driver.
So I'm completely lost for solutions now. I see the following possibilities:
- My WinTV-CI device is borked. This is highly unlikely since WinTV works fine on all 3 test systems.
- The CI driver is borked in combination with the 900H device. Again, then why does WinTV work?
- The CI driver has borked DirectShow support in combination with the 900H device.
The last one is what I'm settling for now, because I noticed that while WinTV is running.... GraphEdit sees no remote graphs to connect to. This leads me to suspect WinTV doesn't use DirectShow internally but just hacks into the drivers directly. And it is the only reason I could think of why WinTV is stable and a generic DirectShow applicaton such as MediaPortal is not. Also, the 900H has specific drivers as mentioned before, so it could be a problem in the 900H DirectShow filters as well.
Anyone see a problem with my reasoning in this?
I've deployed MP 1.1.0 Alpha on my development laptop with Vista Ultimate. The fix I suggested in this topic solves the problem of the WinTV CI device not being succesfully inserted into the graph, however on all my systems (XP, Vista and Win7 now ) the issue remained that it couldn't finish the graph because of the MPEG2 demultiplexer not being connected. I debugged all the graphs being produced, switched some of the pin logic around and that worked fine.
So I got it to the point where my test code scans and detects non-FTA and FTA channels without crashes or errors, which is great. HOWEVER.....
My test code scans 754MHz only band because I know channels are there in my area, while TVServer obviously scans all possible frequencies. If I deploy my custom tvlibrary.dll into the live TVServer I still get a BSOD after 'some time of scanning'. Yesterday after 12 scans it BSOD'd in hwc11bda.sys, which is the CI-driver. After I added some delays in the code between scans, it made some 20 scans before crashing in the same driver.
So I'm completely lost for solutions now. I see the following possibilities:
- My WinTV-CI device is borked. This is highly unlikely since WinTV works fine on all 3 test systems.
- The CI driver is borked in combination with the 900H device. Again, then why does WinTV work?
- The CI driver has borked DirectShow support in combination with the 900H device.
The last one is what I'm settling for now, because I noticed that while WinTV is running.... GraphEdit sees no remote graphs to connect to. This leads me to suspect WinTV doesn't use DirectShow internally but just hacks into the drivers directly. And it is the only reason I could think of why WinTV is stable and a generic DirectShow applicaton such as MediaPortal is not. Also, the 900H has specific drivers as mentioned before, so it could be a problem in the 900H DirectShow filters as well.
Anyone see a problem with my reasoning in this?