Crude AES demo

[MichaelW]

I posted this here, in answer to a question in the Romper Room, to avoid having it get lost.

http://masm32.com/board/index.php?topic=2965.0

[Gunther]

Michael,

nice example. Thank you.  :t

Gunther

[ragdog]

Very nice Michael :t

[jj2007]

Works like a charm :t
... but with a few changes:
if 0   ; ML doesn't like redefinition of symbols...
CALG_AES_128           equ ALG_CLASS_DATA_ENCRYPT or ALG_TYPE_BLOCK or ALG_SID_AES_128
CALG_AES_192           equ ALG_CLASS_DATA_ENCRYPT or ALG_TYPE_BLOCK or ALG_SID_AES_192
CALG_AES_256           equ ALG_CLASS_DATA_ENCRYPT or ALG_TYPE_BLOCK or ALG_SID_AES_256
else
CALG_AES_256x           equ ALG_CLASS_DATA_ENCRYPT or ALG_TYPE_BLOCK or ALG_SID_AES_256
endif
...
    invoke CryptGenKey, hProv, CALG_AES_256x, 0, ADDR hKey

The original ALG_SID_AES_256 chokes with error A2006: undefined symbol : ALG_TYPE_BLOCKorALG_SID_AES_256

I guess you corrected lines 16740ff in WinExtra.inc?

[morgot]

Michael, thank you  :t 

[Siekmanski]

Thanks Michael  :t

[dedndave]

thanks Michael   :t

[MichaelW]

I guess you corrected lines 16740ff in WinExtra.inc?

I'm not on that system right now, but I recall noting the problem with multiple similar equates a few weeks back and adding it to my list of windows.inc errors, but I can't recall if I corrected the lines. In any case I assembled with ML 6.15, no warnings and no errors. Sorry about forgetting to mention this.

I also forgot to call CryptDestroyKey to release the key handle.

[dedndave]

it seems that ML doesn't mind redefinition of EQU symbols, if the constant is the same   
at least, that's what i've observed with 6.15.8803

[jj2007]

I am now on my other system, and see the same oddities in WinExtra.inc, starting with line 16740 and ending with line 16764:

CALG_NO_SIGN                     equ (ALG_CLASS_SIGNATURE or ALG_TYPE_ANY or ALG_SID_ANY)
CALG_RSA_KEYX                    equ (ALG_CLASS_KEY_EXCHANGE or ALG_TYPE_RSAorALG_SID_RSA_ANY)
CALG_DES                         equ (ALG_CLASS_DATA_ENCRYPT or ALG_TYPE_BLOCKorALG_SID_DES)

Do more people have these strange lines, or is it limited to my machines? ;)

[dedndave]

oh oh - i see several such lines, Jochen   :redface:

[MichaelW]

Searching the MASM32 winextra.inc on my second system I find 28 instances of "orALG".

Later:

I'm now on my first system, and I did make the changes to winextra.inc, and CALG_AES_256 is definitely being redefined in my source, so it looks like ML 6.15 tolerates this. I translated the definitions from WinCrypt.h because my first attempts to reference CALG_AES_256, and similar, failed.

[morgot]

MichaelW, can you tell me, how to set specific password? If I do not want to generate a random password, but take a specific string for the password.

[dedndave]

fill in the string with nulls so the user passwords are always the same length
use a fixed key to encrypt the password text
then use the encrypted password data to create the key

just an idea   :P