iTunes HD content not working on my ivy Hackintosh

[wildwillow]

Bit of Googling gives up *some* info, but not much I know...

http://www.tonymacx86.com/mountain-...n7intelgvascheduleavdcommands-tt-err-5-a.html

... indicates it's been seen before. The link in there points to an edit for Intel graphics.

The broken text quoted appears to be Turkish (Google Translate is your friend!) waffling about Open Source, but linked to the Intel internal graphics. Unsure why if you have discrete graphics

One of the posters on the other site mentions a machine definition. I wonder if there lies the answer?

Hi UtterDisbelief,

I have run through all the system definitions to no avail.

This may save you some time#67

 

[wewright]

I'm aware there's no solution yet... which is why I'm still one of the people out there looking more at it.


I've been digging through that driver code a little and found out that this is related to iTunes attempting to utilize the Intel graphics VA API for accelerating the decoding.

A method called CreateAVDAccelerator is invoked which sets one of the data structures to another method (cir address 0x19e88 for anyone else wanting to dig out there). This method has clear indications of being related to decode acceleration (JPEG, MPEG, PP dithering/deinterlacing, etc...). If one of the calls fails (returns with a non-zero error) then it indicates "AVF error: Gen7IntelGVAScheduleAVDCommands, TT err = %d"

I believe the method that returns the error 5 is a subroutine that starts around 0x187ed in the driver. If the value passed into it is non-zero then it defaults to returning 5. It looks like it's always called that way... tedious process but I'm trying to walk it up to see more about what might be happening. Hopefully that'll lead to a more clear view of why it isn't working anymore.

 

[UtterDisbelief]

Hi UtterDisbelief,

I have run through all the system definitions to no avail.

This may save you some time#67

Ah, thanks WildWillow. You have certainly covered the ground with all that testing. I can only say thank you for all that effort!

Ironically, given your results, I would have expected the Mac Mini definitions to have been more tolerant because more than most Apple computers its output can go many different ways (displayport/thunderbolt, DVI, HDMI, VGA) indeed it was designed to be hooked up to any old monitor. The only thing for certain is the Intel HD.

I'm aware there's no solution yet... which is why I'm still one of the people out there looking more at it.


I've been digging ...

... that starts around 0x187ed in the driver. If the value passed into it is non-zero then it defaults to returning 5. It looks like it's always called that way... tedious process but I'm trying to walk it up to see more about what might be happening. Hopefully that'll lead to a more clear view of why it isn't working anymore.

I've always enjoyed a bit of hexadecimal memory/machine-code!

What niggles at the back of my mind about this is the fact that iTunes 11.1.1 on Windows 8, and on the very same hardware, displays all video content as intended.

If iTunes/Quicktime itself was the cause of black-screens etc then the Windows and Mac versions would surely share a common technique to display DRM video as that must be a standard. So why does the Windows set-up work and the Mac one, not?

Perhaps a driver issue? Well I've tried vanilla OS X and nVidia. No joy. If disabling Intel on-board completely in the BIOS will help I'll try it (need to buy a new PS/2 keyboard as USB rarely lets me get into the BIOS set-up any more) BUT it works fine in Windows...

Incidentally non DRM video plays fine. A couple of ones I bought through iTunes were DRM-free and they are unaffected. iTunes/Quicktime is making the choice not to display. So what's the trigger?

It's likely not a memory address, in my view, but a test. If so-and-so then display; if so-and-so do not, etc.

 

[Dekkars]

I'm not sure if anyone has suggested this yet, but this is what worked for me.

1. Downgrade to 10.7, you can find instructions online but I have had zero luck getting iTunes 11 to play any content of mine, let alone HD.

2. Do a CMD-I on the iTunes app and make sure the 'Open in 32-bit mode' is checked.

Hopefully this fixes your issue as it did mine.

 

[wewright]

Hi UtterDisbelief,

I have run through all the system definitions to no avail.

This may save you some time#67

So interestingly enough, there is a reference to a specific board ID in the acceleration code. On the surface it would appear that this board ID is compared to the one pulled from a registry call and different paths are taken if they match. The board ID is "Mac-00BE6ED71E35EB86" which I believe corresponds to iMac 13,1. When you tried the different combinations mentioned in your thread, did you also reset the board ID every time in /Extra/smbios.plist?

 

[wewright]

So no real solutions here, but a few other interesting observations ... in the event someone more familiar with this stuff might build on this for more progress:

I've spend a significant amount of time looking at the AppleIntelHD4000GraphicsVADriver. If you replace it with the version from 10.7.5 or 10.8.2 you get a similar error to that in the beginning of the thread (AVF error: fail to create QT accelerator). 10.8.5 gives you the familiar Gen7IntelGVAScheduleAVDCommands, TT err = 5. Effectively the older driver wasn't getting as far, so I suppose in some poor way, the pink/green is actually better.

The driver from 10.8.2 is less obfuscated than the 10.8.5 one. This file is the native Intel HD 4000 implementation to support the Video Decode Acceleration Framework (https://developer.apple.com/library/mac/technotes/tn2267/_index.html). I suspect that somewhere in version 11.x of iTunes was when the usage of this either became mandatory or was initially implemented (though that's just a guess from the observations of behavior).

The VDA implementation instantiates a data structure with a number of callback methods. The one that the errors are getting thrown from is _Gen7IntelGVAScheduleAVDCommands (or later obfuscated to remove the name but located at 0x19e88 in the 10.8.5 driver). At one point the code attempts to call itself in what appears to be an attempt to schedule the driver delete (i.e., dchedule command _Gen7IntelGVARendererDeleteDriver) and because that fails it emits the TT err = 5.

If you make modifications to skip the error then the behavior is effectively the same ... like the core of the problem is that it's not getting a decrypted frame passed in to accelerate decoding on. You do get these additional messages though:

tid:##### - Mux ID not found in mapping dictionary
tid:##### - Can't handle disconnect with invalid ecid

 

[ajlowndes]

I would love to see a solution to this problem - I have been dealing with this problem for ages, and after ios7 came out simply downgrading to Itunes 10.7 isn't really an option anymore...
Now I just looked in console - I have no idea why I didn't notice that before but I am getting very similar errors to what wewright posted about, especially when I try to scroll backwards/forwards through the video. No favorable result of course, the screen remains blank no matter what. How I wish someone with a boatload of apple engineering knowledge would throw us all a bone - I purchased the damn content after all!

 

[shadywack]

I would think this issue is big enough to defeat a rather significant chunk of the reason to go with a Hackintosh. This not working should really be a front-and-center issue that the entire brain trust focuses on. I'm rather surprised it isn't.

 

[UtterDisbelief]

I would think this issue is big enough to defeat a rather significant chunk of the reason to go with a Hackintosh. This not working should really be a front-and-center issue that the entire brain trust focuses on. I'm rather surprised it isn't.

I totally agree.

What is more, this isn't a very limited occurrence. The Internet is littered with people expressing concern over the problem. Some even with genuine Macs...

I can only assume, rather sadly, that the Apple experts in the "brain trust" (great phrase!) are not suffering from the problem and thus it is escaping notice.

A real shame because, as you point out, not being able to watch iTunes bought content is rather daft.

For me, the ironic and even more ludicrous fact is that on my same machine, but running Windows, iTunes 11.1.2 lets me view the content without any problem at all...

 

[shadywack]

I totally agree.

What is more, this isn't a very limited occurrence. The Internet is littered with people expressing concern over the problem. Some even with genuine Macs...

I can only assume, rather sadly, that the Apple experts in the "brain trust" (great phrase!) are not suffering from the problem and thus it is escaping notice.

A real shame because, as you point out, not being able to watch iTunes bought content is rather daft.

For me, the ironic and even more ludicrous fact is that on my same machine, but running Windows, iTunes 11.1.2 lets me view the content without any problem at all...

Well all we have are suspicions. The console is of no help at all, and all we have so far are behaviors. Like when I enable my Haswell's GPU, but leave my discrete GPU on (an EVGA GTX 570, mobo is a Gigabyte z78 ud5h) I get audio but blank video. Console indicates zilch. Tried all firmwares including new iMac 14,2 definitions and nothing changes. The issue is clearly something to do with the HDCP scheme but without any warnings. The only potential resolution is to have the equipment and engineering know-how to somehow get into the HDCP handshake traffic and see what, if anything is happening there or if it's as simple as Apple locking out external displays. I don't see that as being the situation though given the new Mac Pro's coming out. The number of people on planet Earth with the wherewithall and ability to do such troubleshooting are probably a handful. However some of these same people are the ones who first got Windows installed on an iMac pre-bootcamp and also came up with the Hackintosh foundations. Without their involvement I fear the whole Hackintosh scene is headed for oblivion, and that might be part of what Apple's up to with all this BS.

It's a damn shame. I just don't see myself making much use out of a Hack until the issue is fixed. I'm considering starting a bounty on this problem given it's complete lack of attention from a development side.