Is 7-Zip really a good encryption tool? I wonder what kind of encryption is used in 7-zip. I see most people using 7-Zip. Just curious about what extent it is safe.
$\begingroup$
$\endgroup$
1
-
2$\begingroup$ How about to Google for "What encryption does 7-Zip use?" or similar? I see no effort to find the answer before asking this question. I voted the question down. $\endgroup$– mentallurgDec 16, 2022 at 13:41
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
3
It uses AES-256 with PBKDF2 using SHA1 with only 1,000 iterations as the KDF. The AES-256 is strong, but the password protection is weak so make sure you use good passwords.
-
3$\begingroup$ [THIS COMMENT IS VERY WRONG] Indeed! Addition: the constant is
kNumKeyGenIterationsdefined at line 24 of 7z2201-src/CPP/7zip/Crypto/WzAes.cpp in the source. It's used at line 48. That can't be changed without breaking compatibility with earlier programs, because this is not encoded in the file. The use of the GPU/FPGA/ASIC-friendly PBKDF2-SHA-1, with that low number of rounds, is a dream for password crackers such as hashcat. We need a very strong passphrase to compensate for this, like 6 random words (obligatory XKCD). $\endgroup$– fgrieu-onstrike ♦Dec 16, 2022 at 17:19 -
$\begingroup$ [THIS COMMENT IS STILL WRONG] On second look, no! The 1000 rounds of PBKDF2-SHA-1 are implemented in some crypto supported by 7-zip but NOT for 7-zip's native crypto. I think the correct description is there. $2^{19}$ rounds of SHA-256 is still very far from the state fo the art, but it's 512 times better. $\endgroup$– fgrieu-onstrike ♦Dec 17, 2022 at 11:19
-
2$\begingroup$ On third look, the password-to-key derivation is in
CKeyInfo::CalcKey()of 7z2201-src/CPP/7zip/Crypto/7zAes.cpp there. It hashes with SHA-256 a message consisting of $2^{k}$ times (16 bytes of salt, password, and 8-byte counter), with $k=19$ by default (with parameter $k\le24$ indicated in the archive). So that's about $2^{18}$ rounds of SHA-256 by defaut. Not very good compared to a standard amount of Argon2 or Scrypt, but over a hundred times better than the <$2^{11}$ rounds of SHA-1 corresponding to 1,000 iterations of PBKDF2-SHA-1 $\endgroup$– fgrieu-onstrike ♦Dec 17, 2022 at 11:30