Savedata Encryption/Decryption

[onelight]

It's said that PPSSPP v0.9.6-813 fixed 俺の妹がこんなに可愛いわけがない ポータブル[ULJS00358] savedata bug
   

[sum2012]

It is about these 2 commit

https://github.com/hrydgard/ppsspp/pull/5368
and https://github.com/hrydgard/ppsspp/pull/5369

It's said that PPSSPP v0.9.6-813 fixed 俺の妹がこんなに可愛いわけがない ポータブル[ULJS00358] savedata bug

[popsdeco]

So... I'm reading SED-PC source code and I think
> System.arraycopy(hash_0x20, 0, data, 0x20+0x11d0, 0x10);
must happen just after
> GenerateSavedataHash(data, size, 4, seed);

What I can say so far is that the hashing algorithm is overlapping.
I hope I can tell you more later.

Now at least my sed can also calc correct hash.

[popsdeco]

Hi hykem,

Finally psp-savedata-endecrypter is complete, with help of SED-PC source code.
Also, the following code should give correct hash even for mode4.

http://pastebin.com/DZBFd8jk

[shadow]

is it tested? I can apply it to source since hykem appears away these days

[popsdeco]

shadow,

sadly I don't have any PCs which can execute jpcsp.

umm... I might be able to test using my old Windows, because it has ATI Radeon HD 2400... My recent MacBook is Intel HD 4000 (i.e. out of luck even on Windows bootcamp)

[shadow]

not even in software mode?

[onelight]

Hi hykem,

Finally psp-savedata-endecrypter is complete, with help of SED-PC source code.
Also, the following code should give correct hash even for mode4.

http://pastebin.com/DZBFd8jk

Is this what you say ? http://wololo.net/2014/02/02/release-sed-port-for-pc/
code
https://github.com/hgoel0974/SED-PC

[popsdeco]

onelight, I wrote another implementation based on jpcsp backend:
https://github.com/cielavenir/psp-savedata-endecrypter

But jpcsp's hash calculation code was wrong. So SED-PC author went ahead. That couldn't be helped.
What I can do now is just to compare jpcsp and SED-PC and tell what is wrong to jpcsp team.

[popsdeco]

I tested 3438M, but the calculated hash doesn't match with my hash...

---

Now I'm quite confused. I was able to send a patch for SED-PC to support mode4 hash.
This means both PPSSPP and jpcsp's crypto engines are alright.
So... the problem is lying in KIRK.java?