Strange bytes in DCT4 PPM

[ZeonET]

Hi all!
When I've decompress nokia PPM Text chunks I've found some order of strange (strange for me, 'couse i'm newbie in nokia reversing). Thise bytes like that:
140191C000AF004000BE
140192000072004000BA
1401924000F9004000F3
1401928000A300400009
I found some dependence: there are always such bytes: 14 01 x x x x 00 40 00 x.
When I change them, i can't flash phone (using UFS: write error writing block)

So what mean those bytes? Explain me please.
Best regards...

[ulymar]

writing address.

Regards

[ulymar]

HEX: 14 AAAA AAAA CD00 LLLL CH
AA = Address
CD = Checksum of Data
LL = Length of Data
CH = Checksum of Header

[jockyw2001]

What is the algo to calculate these checksums?

[ZeonET]

hmm... I also tired search algo for this checksums. It's not simple sum (or maybe i make wrong sum). I've check many combinations but no one give a answer...

[indear]

Check TriX source code
http://www.g3gg0.de/index.html?&no_cache=1&tx_eeblog[pointer]=0&tx_eeblog[showUid]=6576

[ZeonET]

Finally! I've solve my question. As indear sad, i look for Trix sourses and found simple algo:
while ( length )
chk += data[--length];
return chk ^ 0xFF;
I think that it must be simple and spin around this algo a long time. I forgot about XORing 0xFF

So demonstrate with an example:
If we have this header: 14 018E 0000 AE00 4000 82
Let's check our header (14 018E 0000 AE00 4000) checksum (0x82)
1. first byte (0x14) and fifth (0x00) must be deleted ('couse they are housekeeping data): 018E 0000 AE 4000
2. Sum : 01+8E+00+00+AE+40+00 == 17D (in HEX)
3. XOR: 1FF XOR FF == 182 -> 82
Oh, it's so nice

Maybe, it'll be usefull for someone else.
Thanks for all for you attention