DecryptHD - Realtime decryption and streaming

[jackelmatador]

Originally Posted by Zotty

Another option would be to make library of this like libcss. I haven't looked at how that one works, but the idea might be a potential possibility.

However just image this would work. How about adding total menu functionality aswell. Would be nice to have complete playback and not only the movie itself. Just pop in the disc, navigate the menu (subtitles, audiotracks, scne selection, extra's, etc.) and have fun for the coming 2 hours.

Don't tease us, that would be an unbelievable achievement!

[jackelmatador]

So I guess my computer is not powerful enough to stream the HD DVD. I just tested for like ten seconds earlier on Thursday, and then tonight I went to actually watch the whole show and nothing but garbage. It seems that my Core 2 Duo E6600, 2 GB of Ram, 7900 GS 256MB Nvidia card, and two 500GB 7200rpm Raid 0 hard drives EXT3 portioning, just doesn't cut it. Is there some option I did not include when I compiled Mplayer? I configured it with xvmc (doesn't matter for this), ssse3, and everything else autodetected.

Here is the error I get from mplayer

************************************************
**** Your system is too SLOW to play this! ****
************************************************

Too many video packets in the buffer: (4096 in 8279842 bytes).
Maybe you are playing a non-interleaved stream/file or the codec failed?
For AVI files, try to force non-interleaved mode with the -ni option.

[qubic]

Originally Posted by jackelmatador

So I guess my computer is not powerful enough to stream the HD DVD. I just tested for like ten seconds earlier on Thursday, and then tonight I went to actually watch the whole show and nothing but garbage. It seems that my Core 2 Duo E6600, 2 GB of Ram, 7900 GS 256MB Nvidia card, and two 500GB 7200rpm Raid 0 hard drives EXT3 portioning, just doesn't cut it. Is there some option I did not include when I compiled Mplayer? I configured it with xvmc (doesn't matter for this), ssse3, and everything else autodetected.

Here is the error I get from mplayer

************************************************
**** Your system is too SLOW to play this! ****
************************************************

Too many video packets in the buffer: (4096 in 8279842 bytes).
Maybe you are playing a non-interleaved stream/file or the codec failed?
For AVI files, try to force non-interleaved mode with the -ni option.

Hi Jackel

I get the same output, but only on some movies. I have a "AMD Sempron 3400+, 2GB ram, nvidia 6200 128mb TC card.
Under windows XP, I got only some sync problems.

regards qub

[Zotty]

I've seen it aswell on a rare occasion, however I haven't been able to put my finger on it. The too many video packets only happens when I play the 2nd part of a movie (e.g. feature_2.evo). So far I've assumed this has something to do with mplayer's demuxing stuff or how the movie was cut in half during authoring. Also noticed that switching demuxer can solve it, but introduces it for the evo that played well with the other demuxer.

One thing I do know for sure is that decrypthd only outputs 2048 at a time, which is exactly 1 packet. Seeing VC-1 playback is relatively new and streaming it like this is apparently the first time ever attempted, these could very well be 'glitches' in mplayer. It's certainly not the first

Try playing with '-demuxer lavf' and see what happens.

qubic, does this mean you've got the openssl part working now?

[Zotty]

Originally Posted by jackelmatador

Don't tease us, that would be an unbelievable achievement!

I've been thinking a bit about this and came to the conclusion this would be a massive undertakement. First of all it would require understanding mplayer's code, which is compact and huge. The decryption part would need to be integrated or provided through a library, including ARF decryption. The disc needs to be accesible through hddvd:// or something like that. So far it's doable.

On a HD DVD disc there are a lot of different parts that need to be processed/parsed. Key files, advanced elements such as images, animations, fonts, music, etc.. And these are all protected. Then you'd also need to parse XML files, make an ECMA script interpreter and put this all together.
Now only the raw material is available and needs to be put to good use. Building menu's and be able to display them with user interaction, the selection of different languages and so on, all integrated with the underlying mplayer code. And there's probably more, which I haven't even thought of.
In any case this would require months of coding if not longer. And don't forget this will be done in spare time since most of us have/need jobs to feed themselves. All in all a very cool idea, but at this time far from being reality.
It would be very nice if there ca be some sort of cooperation between various programmers to realise this. This wouldn't be a simple patch, so this can't be simply dumped into mplayer's dev tree. Probably a seperate source repository would be needed where various people could work on a combined sourcetree. At least to begin with.

[qubic]

Originally Posted by Zotty

qubic, does this mean you've got the openssl part working now?

No. These was decrypted from windows/wine.

--qub

[jackelmatador]

Zotty no worries on the mplayer plugin, that is why I was very surprised.
Hey Qub,
From what I have read the drivers NVIDIA supplies for windows is much better. They supposedly do fun stuff like actually sending VC-1 frames to the video card to be processed. I know in Linux only MPEG1/2 is done this way through XvMC.

I think once I figure out how to make the video play acceptably (-lavf is a little better) I might write a little script that just plays the biggest EVO files available. The only tough part would be to make sure they play in the right order, for now it seems all movies have some sort of numerical suffix.

[XAvAX]

Perhaps a way to implement it without requiring root permissions would be to make a daemon "backend" that runs as root and reads the files, only to pass them (without preprocessing, so that the userpriv part can be reniced individually) to a "client" running with user privileges, which does the actual decryption? In theory, this could also enable:

Multiple users accessing the disc
No need to be root
Individual config files (default drive location, fallback VUK's if implemented)

Maybe, eventually, we can wrangle someone who understands FUSE to implement a sort of "pseudo-filesystem", like eCryptFS, which acesses the mountpoint, and creates a sort of "translated" filesystem in a directory that decrypts files dynamically on access - this is essentially what eCryptFS does, so it's feasible, and it might even be possible with existing code since AACS uses AES and there are already several programs, including this one, which can handle subset-diference. In theory, this could make that menus and extras concept possible.

Whap me over the head if I'm being too blue-sky, but does anyone know a filesystems person?

Also, I found a post on vger.kernel.org talking about adding a uid=ignore/gid=ignore option to mount for UDF, dated to March 4 2006. Try it and see if it helps.

[unixfs]

I didn't test this software because I don't have a drive, anyway I have a couple of hints and requests:
- adding -demuxer mpegps isn't necessary (it will only detect the mux format faster) and as already observed by others it isn't even the best option available
- -demuxer lavf -correct-pts may playback smoother
- hddvd:// would require a stream reader module in mplayer (look at stream/ directory) which in turn would require a reader library that all other softwares may use, such as libdvdread that provides access to single titles without bothering with menus, stills and so on
- playing single files _will_ break multi-angle movies, just like playing single VOB of multi-angle dvds _does_ break today, thus a library is the only way to go for the future.

Now the requests, solicited by you: may you release the source code, please? Possibly to some plain ftp/http server (e.g. not rapidshare and co).
Thanks very much for your work!

[Zotty]

A little later than planned (was busier than expected), but here it is;
The source: decrypthd_src_20070611.rar
And the binary: decrypthd_bin_20070611.rar

Changelog:
- Simultanious save to file while watching (use <output dir> -o)
- CMAC AES (for Volume ID verification)
- Code cleanup
- Resets AGIDs so it's no longer needed to manually reset the drive

[Zotty]

Originally Posted by XAvAX

Also, I found a post on vger.kernel.org talking about adding a uid=ignore/gid=ignore option to mount for UDF, dated to March 4 2006. Try it and see if it helps.

Tried your suggestion, but unfortunately no change when adding adding it to fstab (user,defaults,uid=ignore,gid=ignore).

Originally Posted by unixfs

- adding -demuxer mpegps isn't necessary (it will only detect the mux format faster) and as already observed by others it isn't even the best option available

During playback of files from harddisk this is indeed not necessary as autodetection works fine. But it doesn't work when streaming. Don't know why, but without demuxer it just outputs some text, but no video. It seems detection doesn't work properly when streaming.

- playing single files _will_ break multi-angle movies, just like playing single VOB of multi-angle dvds _does_ break today, thus a library is the only way to go for the future.

Are there any multi-angle HD DVD's out there? Haven't seen any yet.

Now the requests, solicited by you: may you release the source code, please? Possibly to some plain ftp/http server (e.g. not rapidshare and co).
Thanks very much for your work!

See my previous post Unfortunately I've only got webhosting that's linked to my real name. And I'd rather not use that one due to potential 'trouble'.

[qubic]

Originally Posted by Zotty

A little later than planned (was busier than expected), but here it is;
The source: decrypthd_src_20070611.rar
And the binary: decrypthd_bin_20070611.rar

Changelog:
- Simultanious save to file while watching (use <output dir> -o)
- CMAC AES (for Volume ID verification)
- Code cleanup
- Resets AGIDs so it's no longer needed to manually reset the drive

Hi Zotty

This is great. But I get this fare:

[root@dhcppc0 decryptHD]# make
g++ -c -o src/decrypthd.o src/decrypthd.cpp -I.
src/utils/file.hpp:6: error: conflicting declaration ‘typedef long long unsigned int uint64_t’
/usr/include/stdint.h:56: error: ‘uint64_t’ has a previous declaration as ‘typedef long unsigned int uint64_t’
make: *** [src/decrypthd.o] Error 1
[root@dhcppc0 decryptHD]#

openssl was configured with this:

./config shared --prefix=/usr -BN_LLONG

regards qub

[Zotty]

The typedef in file.hpp is no longer needed, due to stdint being used now.

Try replacing:

#include <string>

typedef unsigned long long uint64_t;

class File

with:

#include <string>
#include <stdint.h>

class File

That should solve the conflicting declaration. Will also fix this in the code over here.

[BENtheTEN]

Hi there

I've got a problem using decrypthd. I'm always getting the following message in the log:

*** Starting decryption of PEVOB_1.EVO ***
Dumping to stdout
Getting key number from /media/CONSTANTINE_HD/HVDVD_TS/PEVOB_1.EVO
Needed key is
No keynumber. Aborting!

### Exception: HDDecrypter: No keynumber. Aborting! ###

I tried the source and binary version.
Is there anything I do wrong? Am I supposed to put a Key somewhere?

My System is 32bit Debian SID, external xbox drive

Thanks
Ben

PS: dumphd worked flawlessly with all my HD DVDs using exactly this drive

[BENtheTEN]

Hi
I "solved" my problem by replacing

ifEvobFile.read((char*)u8Header, sizeof(u8Header));

return u8Header[0x3E] - 1;	// note: we start at 0, while keynumber starts at 1

with

ifEvobFile.read((char*)u8Header, sizeof(u8Header));

return u8Header[0x3E];	// note: we start at 0, while keynumber starts at 1

but I have no idea why it works.

Regards
Ben

[BENtheTEN]

hi

did I miss something or is there really no improved decrypdhd version? like watching blu-rays on the fly.
I'm not happy ripping every disk and watching it this way. It's very time and disk space intensive.
Unfortunately I'm not a (good) coder, so I can't improve it by myself.


Best Regards
Ben

[pardus]

Originally Posted by Zotty

A little later than planned (was busier than expected), but here it is;
The source: decrypthd_src_20070611.rar
And the binary: decrypthd_bin_20070611.rar

Changelog:
- Simultanious save to file while watching (use <output dir> -o)
- CMAC AES (for Volume ID verification)
- Code cleanup
- Resets AGIDs so it's no longer needed to manually reset the drive

Would you mind opening a Google Code project so we could get the source easily. Sendspace links are dead now. Thanks.

[pensivepuppy]

Can someone re-post the source code, or send me a copy. The sendspace links are no longer valid.

[bshep]

for the permissions problem you could try the umask= option.

If I remember correctly if you set it to umask=000 all files will have full permisions (rwxrwxrwx).

I think this only works if the FS has no specific permsions, not sure how UDF is structured so it may or may not work.

[v1rtu0s1ty]

I should have not bought the Toshiba HD-A2 player. I should have just saved more money for the xbox hddvd drive.

Awesome tool! I will try it one of this days!