,
9 tweets,
6 min read
Read on Twitter
So I wanted to encrypt some files. Thought about using 7z+password. Stackexchange folks said "Didn't review it but it should be fine. You can browse the code yourself". So I did. After a few mins I noticed they use 8byte "random" IV. Yes, half of IV is zeroes. But it gets worse.
Aside from the fact that 7z XORs plaintext with zeroes, I was curious about the "RandomGenerator" which generates IV. That's when I vomited. The first comments say "This is not very good random number generator. Please use it only for salt." It is not used only for salt.
Yes, it uses PID and time(null) as seed. Yes, this is 7zip's random generation code running on your computers in 2019. More "cursed" code below. #7zip #encryption #facepalm #randomness #entropy #RNG #crypto #cryptofail 
Now that was just the Init() of RandomGenerator which happens once, the next random values are generated as shown here. Yep, it mixes in a totally random hardcoded salt to previous SHA256 state. Then returns up to 16 "random bytes". Doesn't check that 'size' argument < 16.
And there you have it folks. Open-source "many eyes have looked at it for years so it must be secure" crypto code. AES256-CBC but supplied with IVs half of which are zeroes, and the other generated by a toy RNG. That's just from a 30min review. #7zip #7z #opensource #FOSS #crypto
If you think it's better to use zip encryption instead of 7z in 7zip, here's a recent @whitequark tweet to prove you wrong:
I thought about reporting this at 7zip Sourceforge forums but then I vomited again when I saw a long thread of largely incoherent exchanges on how 7z should be using Twofish instead of AES-256 because obviously NSA backdoored AES back in 2001 didn't you hear
Seems typical of crypto/security code reviews in general - a bunch of folks fighting over which hipster encryption/hashing/signing algos to use, while overlooking the most obvious vulns and holes visible to anyone with half a brain and a few mins to spare. End of rant. #security
Reported the bug here: sourceforge.net/p/sevenzip/bug…
