off-topic but it's really mind-bending how much money google must make on ads in order to justify serving all this stuff for free and still making huge profit
#cybersecurity
grizzled lake
tight abyss
doubt
as long as some parts are highly profitable, others can make loss and just exist for PR, basically
I'd guess most money comes in from enterprise cloud business and stuff like that, which carries free public services like youtube
grizzled lake
that too. but there was a time where they werent making money on cloud and enterprise
or at least not as much as now
theyre public, i guess one could go thru their annual reports and see
thorn obsidian
@grizzled lake I use Firefox
Are you talking 3 "Find the item"s? Because yes, I do that as well. I thought you meant more as in, you'd complete a captcha and you'd have to do another one after you've sent the last one
Poor understanding on my part 😄
late hound
Thought it would be of interest in this channel.
thorn obsidian
@late hound Has this been fixed?
late hound
I have no idea, the reddit post along with the article is hinting to me that it hasn't been fixed yet.
latent kelp
Then we don’t allow such discussion here, thank you for understanding
late hound
Oh my bad, I just thought it was an interesting and important piece of news.
latent kelp
It is, but until it’s been patched and rolled out there’s still a potential security risk and we don’t want to be part of that
late hound
Of course, I understand. I wouldn't want to be messing with Valve either haha
arctic vale
Probably never gonna be fixed, unless all the attention forces them to fix it. They have several similar CVEs dating back years that are still working.
late hound
Actually? I didn't know that they just didn't care at all...
cedar pelican
@thorn obsidian Sorry to bother you again. I changed my mind on sharpops being able to be connected externally. I'm making it so you can only trigger updates to already running services.
But I found this today:
# Re-direct to remote environment.
export DOCKER_HOST="ssh://my-user@remote-host"
# Run your docker-compose commands.
docker-compose pull
docker-compose down
docker-compose up
# All docker-compose commands here will be run on remote-host.
# Switch back to your local environment.
unset DOCKER_HOST
Is this a unsecure idea?
The ssh key would be hidden behind circleci in my case.
thorn obsidian
It's not something I would do, personally.
cedar pelican
What reasons do you have?
I want you to be able to deploy sharpops a bit like how you deploy heroku
But I can't see how I can make it secure
thorn obsidian
I wouldn't use anything through root
A single mistake and the system is hosed at that point
cedar pelican
I don't know how I can avoid it and give the functionality I want
Docker being the main problem
The docker dameon is root
Do you not think that is a problem?
thorn obsidian
I think that having a server send commands to root is questionable.
cedar pelican
And just a quick question: if I was to make it its own sharpops user, do I start the service as root, and they it "changes" itself down to sharpops user?
Cause i want to make that process seemless
So just a os.system(su sharpops)
thorn obsidian
Well, I'd hope you wouldn't use os.system() for one.
thorn obsidian
Oh, I have no idea the proper way of execution then if it's not Python.
You mean like how Apache/Nginx/etc works?
cedar pelican
I don't want the user to have to do sudo su sharpops
Wait
I use a systemd service
I can just make it non-root
can't I?
Is that sutiable?
thorn obsidian
Try it out and see
cedar pelican
Cause then its just a case of dealing with this
I might just analyse compose files before they run
And if its trying mount a volume to host
Error
But that's again relying on my filter programming skills
@thorn obsidian If it is a non-root user, I think I can accept that needs to happen
What other security things Do i have to think of?
thorn obsidian
Proper escaping of SQL, XSS, CSRF, just to name a few
cedar pelican
Theres no SQL anywhere
XSS is a issue
Im not going to allow you to run commands remotely#
I am going to allow for a docker-compose file to be downloaded and run
So i need to add filters to that
For example, you can specify what locations it can download from
For example, only your github
grizzled lake
@cedar pelican apparently nginx works like that, where it starts as root and then de-privileges itself to run as the nginx user
no idea how that works but its possible
grizzled lake
yeah you can tell systemd to run as a specific user, if you don't need it to start as root
take heed
thorn obsidian
you can specify the runas-user in systemd units, and you can also deescalate privileges via setgid() and setuid()
thorn obsidian
additionally, umasking and chrooting are also good practices after dropping privs
sacred elm
anyone wanna try solve my cipher for the memes
finite idol
yes
sacred elm
AFDDGGFXGFDDFADAADXAAAFAGDGFDADDXGXX GXFDDAXGAGDX GGAFXDAGGD
@finite idol lol have fun
quiet viper
So that's interesting
sacred elm
yeah @quiet viper i upgraded it again
and was a bit nicer by giving a hint in the solvable part
quiet viper
Oh what I was actually going to say was that your cipher triggered our token alarm
sacred elm
bahahaha
finite idol
lmaoo
quiet viper
Gave me a bit of a chuckle
finite idol
how does the token alarm work
sacred elm
yeah mayo, if you can crack that bad boy then you are a true god
quiet viper
I'm thinking it's just a regex
silent pier
Mostly cause the word fag is in it
sacred elm
yeah i noticed that part Floppy, gave me a good chuckle
quiet viper
Oooooooo THAT regex alarm
silent pier
:P
silent pier
Would delete it then
quiet viper
Yeah I was wondering
finite idol
I know nothing about cryptography so i probably wont be able to solve it, but can you tell me how many characters long the output will be?
sacred elm
if any of you can solve that bad boy and tell me the pass phrase ill give ya $100
finite idol
also for people to be able to solve it and figure out what the passphrase is, you'd have to tell them what the pass phrase does / how the algorithm works i think
sacred elm
the output is 18 not inc spaces
finite idol
ok
sacred elm
well ill give you this, the two pass keys can be solved through 2 ciphers
the actual phrase needs 3 but you need to do them in the right order and you need to do them the correct way
hence why i said my cipher was a big boy secure one
and i have to do it by pen and paper because no software can encode/decode using my custom method. That is untill i can work out a way to write a script to do it for me
AFDDGGFXGFDDFADAADXAAAFAGDGFDADDXGXX GXFDDAXGAGDX GGAFXDAGGD
those two at the end are the pass keys
thats another clue
hence why they are seperate from the rest of the cipher base
finite idol
oh i thought those were seperate words that would be decoded
sacred elm
nope
the actual pass phrase words are all joined together and only get separated once you have solved the 3rd cipher type
ill give you another clue
AFDDGGFXGFDDFADAADXAAAFAGDGFDADDXGXX GXFDDAXGAGDX GGAFXDAGGD
that is 4 words
finite idol
so the answer is an english string?
finite idol
ok cool
those are short words
4.5 letters on avg
actually thats about average for english i think
sacred elm
this wouldnt be solvable in another language unless the language shared the same alphabetical characters with no fancy ones
sacred elm
yeah mayo the key is to keep the words short and limit words with repetitive letters, it makes it harder to brute force through the 3rd cipher
finite idol
cool
sacred elm
harder than it already is considering i eliminated the pattern factor
i could make the cipher even tougher to solve which i might end up doing in the future but by that stage i would only do that if i made myself an executable to encode and decode for me
because its already hard enough to do by paper, takes 5-10 mins
another trick is to use a traditional cipher but mix up the way you use it which is what ive done for 2 of the 3 ciphers, means that when brute forcing its harder because you wont always know what exactly you are trying to brute force and if the result it pumps out will be correct or not
as you might be able to tell i love puzzles, hence why i want to learn python and get into the web development/cryptography side of things
sacred elm
lol have fun with that
finite idol
ty
sacred elm
im not even sure it would be possible to brute force this (knowing my luck it will be easy as heck to brute force)
finite idol
oh its possible
sacred elm
id be interested to see if it is/how long it takes just so i know for future
finite idol
i think that no matter how long the string, the absolute maximum possibility of attempts it will take with my method is 4.0329146e+26
but im bad at this stuff so theres probably a way better method
gentle heron
cracking a code with so few samples is pretty hard
compared to that finding a fast way to bypass it once you know the algo is relatively easy
finite idol
yeah
sacred elm
true
finite idol
if you gave me a 100 character string i'd be done i think
sacred elm
i mean i could do that, the question is do i have the effort or time to handwrite that out
gentle heron
if you can do the encryption by hand you can get a program to do it
finite idol
that's what Sully came here to learn how to do lol
gentle heron
icic
sacred elm
@gentle heron the pass phrase is 4 words that are apart of the english language and can be found in the english dictionary and are comprised in a proper sentence
gentle heron
yeah but thats not really much info about the algo
sacred elm
yeah im learning python to make an encoder/decoder for my cipher
finite idol
he gave some info on the alg earlier
sacred elm
it is 3 cipher types that are all personally customised to through people off
finite idol
can it throw computers off though?
sacred elm
last two strings are the passkeys which are only 2 ciphers to solve but now that i think about it i used the pass key of one of the ciphers to cipher the passkey for that cipher fug
gentle heron
in order to figure out the cipher we would need some examples that include the plaintext tbh
right now its just a bunch of random characters
sacred elm
i mean one cipher seems pretty obvious by looking at it
finite idol
with a long enough sample i was planning on mapping each two character pair to a random character and doing a character frequency analysis to convert that to english
but we only have 18 character output
sacred elm
A, D , F , G, X ....... that means one of the ciphers is an ADFGX cipher
the thing is i ciphered the passkey for that cipher using that cipher so i kinda messed up there
so pretty much just exclude the last two strings and thats all you are able to solve, you will just have to try work out the passkeys using luck
AFDDGGFXGFDDFADAADXAAAFAGDGFDADDXGXX
thonks that maybe i can use a new cipher to cipher the passkeys
latent kelp
you’ve posted that code four times already, for the next one please just link to a previous one rather than pinging every moderator again thanks
sacred elm
sorry @latent kelp
finite idol
lmao the alarms
dawn bluff
Where do I start with security @ me if you have some good resources
flat creek
Really depends on where you are at, programming wise. Look into encryption, OWASP's top 10 vulnerabilities, certificates, google any vulnerabiliies you hear about, practice with websites like hackthissite, hackme etc, practice with web applications like bWAPP, "damn vulnerable application" or what's its face, that should get you going @dawn bluff
dawn bluff
Ok
fallen ginkgo
Is the Michael Bazzell OSINT book still valid
upbeat palm
Never heard of it, look at tracelabs.
thorn obsidian
192.168.1.6 ba:33:2b:87:e1:16 (Unknown)
192.168.1.8 ba:33:2b:87:e1:16 (Unknown)
Is this possible? 1 device with 2 IP Addresses?
wanton rune
Yes, if you have more than 1 ethernet driver
thorn obsidian
I see
fallen ginkgo
Interesting thx robin
gentle heron
you dont need more tha none driver, in most operating systems you can just straight up assign as many ips as you want @thorn obsidian @wanton rune
I use it all the time at work since I might need to be able to access multiple subnets in a few minutes
Eg when I give new devices an IP itll almost always be on a different subnet than the default one so I just put in an IP on both subnets so I dont have to change any settings while im working
thorn obsidian
I see
gentle heron
at one point i had 8 ips on my work laptop
thorn obsidian
What teh
gentle heron
lol I had a bunch of jobs going on so i left the ip i needed for each site in so i didnt have to mess with it
thorn obsidian
Did a scan on those ips, found out they were running netbios
gentle heron
sounds like windows machines then
thorn obsidian
Yeah that's what I'm wondering. I'm dual booting windows and linux. But I rarely use windows. And there's only one computer here
So I'm trying to figure out where those 2 are coming from
The MAC vendor can't be identified
:/
gentle heron
is that your computers mac address?
thorn obsidian
Nop.
gentle heron
i assume those are from dhcp, it might have given the router a hostname
i often use wireshark to try to identify systems
gentle heron
it also can tell you the manufacturer based on a lot of mac addresses
gentle heron
yeah its nice
thorn obsidian
I took the mac address online, none of them could give me any results
gentle heron
i use it to identify our hardware when they didnt let me pre-configure devices and now i need to know what stuff is actually installed and whats not working
or to ID stuff thats existing that i need to move over to the new one
eg like I know we are installing samsung cameras but I see some axis already there
thorn obsidian
Hm. I see.
gentle heron
and wireshark knows their mac address range 👍
or when the guys dont WRITE DOWN THE MAC ADDRESSES 😡
so i gotta find it
thorn obsidian
I was also wondering, what are TCP fingerprints?
See them in nmap scans
Can't understand them
gentle heron
um, sounds like when you look at how a system builds its packets and use that to identify the operating system
like some operating systems would use sequential packet ids
so you could watch for that
thorn obsidian
I see
gentle heron
now how does it work in nmap? sorry id have to look it up and im working on some other stuff and dont have room in my brain for that right this sec lol
thorn obsidian
lol, that's ok
gentle heron
check out their big tutorial pages though https://nmap.org/book/man-os-detection.html
gentle heron
if you are on windows :
control panel-> network and sharing center-> the interface you want to edit-> properties-> IP v 4 properties ->advanced
lol
huge pain in the butt
sacred elm
Anyone have any scam emails sent to them recently asking for you to sign into something?
noble kraken
nope
subtle parrot
I have a program to search a specific WIKIA for information and do some filtering with them then display the info to user. Except I have to login as a bot through username/password which I do not want to hard-code per common sense. How should I store the credential?
Pack it in a file somewhere, load the file at runtime and put the file it .gitignore?
stark vortex
most operating systems these days have systems to securely store user credentials and ask for them at runtime, requiring user confirmation
silent pier
How hard is it to implement a simple authentication system for 1 route on a web server?
tight abyss
you could just slap some nginx with http basic auth in front of it and be done 🤷
silent pier
The flask server is using nginx
but I'm running sapper for the frontend, which needs the auth
cedar pelican
Lemme give you a hand then
@silent pier You choose the chat, mention me and I'll help you out :D
silent pier
I'll hit you up tomorrow, I'm currently trying to bore myself to sleep @cedar pelican
cedar pelican
Hahahha ok
stable flower
Any way to fingerprint what framework does the website running on nginx use?
Don't Angular, Django or React for instance leave any sort of traces on the source code of the page itself?
cedar pelican
I mean, you could maybe guess by reading the HTML code to a point
Each framework will have a "trademark" printing layout etc
This is more of a question: is that actually a valid way of finding something like this?
stable flower
nvm, apparently https://builtwith.com was what i needed
silent pier
I know there's an addon that does what you want
- lists all tools, might just be that one from all i know
thorn obsidian
@sacred elm I was getting spoofed e-mails saying they had hacked my account
Which is trivial to get rid of if you verify the Return-Path is the same as the From header if it says it's from your own account.
Which, since implementing that, funnily enough, I've gotten no e-mails saying they've hacked my account 😄
The funny thing here is that spoofing the From header is trivial to do. Maybe 10~ lines total to send a spoofed e-mail, if not less. Which is why verification of headers is so important.
sacred elm
@thorn obsidian I was getting the same thing but it was “Apple” telling me my account has been disabled and I needed to verify my details to re-enable it. I got the email 6 times in a week but I dealt with them by running my script to send them fake emails and passwords and confuse them on which details were legit
thorn obsidian
Eh... I wouldn't do that.
sacred elm
The easiest way to tell it’s a scam is by reading the email address
silent pier
Common sense is the biggest tool in revealing scams
sacred elm
true
thorn obsidian
Any level of "hacking back" is wrong, and at worst, illegal.
sacred elm
It’s not really hacking, it’s just sending them a heap of fake emails and passwords. I could do the same thing without a script it would just take me longer
thorn obsidian
I read it wrong because of the script part.
sacred elm
ahaha all g
safe bear
thorn obsidian
Neat
finite idol
hey guys, I have a question / challenge. given the /etc/passwd and /etc/shadow files of a machine, how would you go about breaking into said machine? is it even feasible?
here are some example files (they're not anyone's actual /etc/passwd or /etc/shadow files) if anyone's interested
thorn obsidian
with these? you can't use them to break into the machine. You can try throwing the hash into a cracker but if it's strong enough, you will not crack it in any feasible timespan, considering it's a sha-512 hash
finite idol
cool, thanks
simple orchid
googling part of that hash shows it's part of a CTF and apparently a weak easily cracked password according to this writeup https://github.com/mzfr/ctf-writeups/tree/master/picoCTF-2018/Cryptography/ HEEEEEEERE'S Johnny!
so basically doing exactly what xx said will work, but don't expect it to work in the real world with strong passwords
finite idol
That's neat, i've never heard of the John the Ripper tool before
thorn obsidian
@simple orchid good catch! should've googled the hash :p
i've actually even done that exact same picoctf task
Of course, that's only implying there's a password between you and login.
That doesn't allow you to break into, say, FDE or 2FA ( https://github.com/google/google-authenticator-libpam Google Authenticator PAM for example )
mellow steeple
can anyone imagine what kind of hash is this 2272fc5d5c4a170baf20bbe7f489e09ad74627c1
or what is the best way to find algorithm of unknown hashes?
tight abyss
look at the length
you have 40 characters in hex notation, means 20 bytes, means 160 bits
That's usually SHA1
actually you can just throw it into duckduckgo and it will tell you what kind it is - plus a wikipedia link 😄
even google doesn't do that
@mellow steeple
thorn obsidian
@mellow steeple What's this from?
mellow steeple
from a online game they use this kin of hashes to encrypt there links to game assets
thorn obsidian
Which game?
mellow steeple
not a popular one and its {deleted}
thorn obsidian
What's the name?
mellow steeple
{deleted} do you know?
thorn obsidian
I don't, but I'm curious where you got the hash from. Can you link it?
mellow steeple
why tbh? (i'll send)
{deleted}
this is an old one
did copied it?
i have to remove this....
thorn obsidian
I didn't get it, no.
thorn obsidian
Wait, that's not what you think it is.
mellow steeple
yep its just a link there is another part after that hash its the assest
sorry but i think its illegal sharing them...
if u really really want it let e know
thorn obsidian
So, question
If you download the file that links to, and check it's sha-1 hash - does it, by chance, equal 2272fc5d5c4a170baf20bbe7f489e09ad74627c1 ?
Because I think you believe it's some kind of encryption, when it's probably just the sha-1 hash of that file.
Which is why it comes up as a sha-1 hash
Sort of a strange way to do the check, but I've certainly seen worse.
mellow steeple
they use base64 to encrypt their files but these just links ( links to the files )
thorn obsidian
base64 isn't encryption
mellow steeple
oh...yep sorry
thorn obsidian
you run sha1sum on the file, and it should come up with the same hash as it was linked with
mellow steeple
ok here is one of files if anyone interested http://assets.millennium-war.net/9b53a542a0a8a852326c198e02f91f587c5b5065/1fp32igvpoxnb521p9dqypak5cal0xv0
thorn obsidian
834c19cb19981796a3b1cbfa683dfde78474140c is the hash, so no clue.
I'm not in the mood to binwalk the file
high entropy
i think it actually might be encrypted
mellow steeple
isnt 9b53a542a0a8a852326c198e02f91f587c5b5065 a hash then?
thorn obsidian
it's definitely not base64 as there are no ascii characters
and i was worng, entropy was low
mellow steeple
what is that?
thorn obsidian
a visualisation of the entropy of the file
mellow steeple
i really dont know tbh
thorn obsidian
there are no 0x00 bytes, no 0xff, no ascii bytes, no low bytes
mellow steeple
is there any way to decrypt hashes w/o brute force?
mellow steeple
...
time to give up current method and try another method...
thnx
btw could you decrypt that file?
thorn obsidian
no, but there's tonnes of interesting repeating patterns
i'm not sure what the hell this is
mellow steeple
can u explain what are those blue and black backgrounded text means if u have enough time plz?
thorn obsidian
that's just converting it to characters i can display in my terminal
some of the characters still can't be displayed, which my pager program less then displayes as that blue highlighted unicode
mellow steeple
there was a script that could decode those files someone have made...
mellow steeple
there is a link that gets updated every time game updates and its really a file list
that link is also a encrypted one
i needed to make a script to trace that link
that link is showed when we send a get request ( not us the game )
first part and last part are same but middle part get changed of the link
thorn obsidian
just reverse engineer the thing that uses it
mellow steeple
probably yes.... if i could check each get request and the thing that does when i pressed the start Butten of game ..... thank u very much for the idea
spiral turtle
So anyone got any code that detects or scrapes proxy servers?
thorn obsidian
oh
@spiral turtle you mean socks sorta proxies
funnily enough, i do have one, i will send it in dms
spiral turtle
gg
tight abyss
rip
thorn obsidian
any ideas how an ability to sign up without providing the password at all can be exploited? To my understanding, blank string can also be hashed, so flaw doesn't make big difference hash wise.
a blank password isn't inherently a security issue implementation wise, it's just a dumb idea to let users not input any password
and @ your hash comment, any platform worth a damn also salts the password beforehand, so in this scenario the entirety of the password would be the salt
@thorn obsidian salt can be random and not random right, I've heard there were 2 different approaches, but in case i am aware of what the libraries the web server uses, is there any chance that the prevalent modules in those libraries that are related to salt/hashing could be worth looking into?
they're never "random", you need to be able to reproduce the salt when you're checking password hashes
they're randomly generated, sure, but they can be reproduced
i see
often, the salts are saved in the database in plaintext
what if the server tends to disclose exceptions of the library itself on the webpage, including variable names and such, is there any way to provide such password that would cause an error
or is completely based on sanitization on the code side
are there any fundamental character blacklists defined by hashing algos? Or am i allowed to use any symbol in any encoding
algos don't hash characters, they hash bytes
otherwise, they couldn't hash eg files
to be technical, they hash bits
but for all intents and purposes, you don't need to know that
so yeah, it's totally up to the sanitisation of the site itself what it decides to feed into the hashing algo
thanks
np dude
cedar pelican
Hi.
So I had my first case of someone registering an account using a email from my domain.
I emailed the company asking for the account to be closed, but basically I wanted to check. Am I responsible for what happens on accounts like these?
spiral turtle
yes
spiral turtle
I need help with reading this data
from the response
I need to pick out each bit without all the brackets then convert it to a dictionary
but idk how to remove the brackets
Target IP: 104.27.200.92
Target Port: 443
{'OCSP': ('http://ocsp.comodoca4.com',),
'caIssuers': ('http://crt.comodoca4.com/COMODOECCDomainValidationSecureServerCA2.crt',),
'crlDistributionPoints': ('http://crl.comodoca4.com/COMODOECCDomainValidationSecureServerCA2.crl',),
'issuer': ((('countryName', 'GB'),),
(('stateOrProvinceName', 'Greater Manchester'),),
(('localityName', 'Salford'),),
(('organizationName', 'COMODO CA Limited'),),
(('commonName',
'COMODO ECC Domain Validation Secure Server CA 2'),)),
'notAfter': 'Nov 7 23:59:59 2019 GMT',
'notBefore': 'May 1 00:00:00 2019 GMT',
'serialNumber': '8B298ABE2131F3A53B8E55F6A667AF45',
'subject': ((('organizationalUnitName', 'Domain Control Validated'),),
(('organizationalUnitName', 'PositiveSSL Multi-Domain'),),
(('commonName', 'ssl382558.cloudflaressl.com'),)),
'subjectAltName': (('DNS', 'ssl382558.cloudflaressl.com'),
('DNS', '*.luxelifescape.com'),
('DNS', '*.mediabase.ga'),
('DNS', '*.openmany.tk'),
('DNS', '*.plusone8.com'),
('DNS', '*.spigotmc.org'),
('DNS', 'luxelifescape.com'),
('DNS', 'mediabase.ga'),
('DNS', 'openmany.tk'),
('DNS', 'plusone8.com'),
('DNS', 'spigotmc.org')),
'version': 3}
TLSv1.2```@spiral turtle Just make it a string, remove the brackets
spiral turtle
mk
cedar pelican
from ast import literal_eval
python_dict = literal_eval("{'a': 1}")
Then do this
@spiral turtle
spiral turtle
k
cedar pelican
Can I make the perms of the docker-compose executable 775 safely?
If docker itself is rooted?
tight abyss
why would you?
tight abyss
why executable then though?
cedar pelican
Without using a unix group
User 2 is a automated user
So it can't sudo
Do I just chmod it in its name?
And sudo for other users?
tight abyss
it could, if you configure sudo appropriately
or that, make sure always the same user is used
I still don't see what anything of that would have to do with executable/775 tho
cedar pelican
well 700 would mean only the owner could execute it
755 is what a executable should be in theory
But I need 775 for my usecase
Or 755 with user 2 being the owner, and user 1 using sudo
tight abyss
but you never execute a yaml file
or am I understanding this all wrong
OH, you mean the actual executable command file, not a docker-compose.yml
tight abyss
7 means writable though
cedar pelican
Well 7 means all 3
tight abyss
4=read, 2=write, 1=execute
tight abyss
so with 755 everyone can run it anyway
nobody (except root) should be able to modify executables run by others
so... I'm only more confused now, I think
thorn obsidian
@cedar pelican Which site did they register with?
So I had my first case of someone registering an account using a email from my domain.
cedar pelican
@thorn obsidian There is email verification.
And by random string as a email I mean euaisnidnsjajs@mcaq.me
But my point is a lot of companies let you use the account even if the email ain't verified
thorn obsidian
Huh. That's interesting
I don't really have a recommendation outside of not having catch-all addresses/only allowing mail to specific addresses, but that doesn't really solve the problem.
proper magnet
thorn obsidian
Looks like a substitution cipher
I'm not sure if we have anyone trained/qualified for cryptography/encryption here. I'd recommend #crypto on Freenode for that
cedar pelican
@thorn obsidian Again, not all companies do a "email verification" check, even to check if it's real in my case
So having a catch-all is useful for me to deal with events like this
But do you think I can still safely remove the catch-all?
cedar pelican
Ok so i'm trying to make a url filter : it will only be true if a url matches a filter url
Here is the code:
package main
import (
"strings"
"fmt"
)
func main() {
string1 := "raw.githubusercontent.com/Sharpz7"
string2 := "https://raw.githubusercontent.com/Sharpz7/mcadesigns/master/docker-compose.yml"
string2 = strings.ReplaceAll(string2, "https://", "")
string2 = strings.ReplaceAll(string2, "http://", "")
string2 = string(string2[:len(string1)])
fmt.Println(string2)
if string1 == string2 {
fmt.Println("Correct URL")
} else {
fmt.Println("WRONG LOCATION ERROR ERROR")
}
}
orchid notch
sharp zis is not your golang server :p
cedar pelican
This is its true position, where string2, the user data, is what you edit
And string1 is the filter
Is this a secure way of doing it? Is there any weird internet tricks to bypass this?
thorn obsidian
Huh, removing locksettings.db from /data/system disables the lock screen in Android Oreo. TIL
Pre-8, just remove the .key files from /data/system 😄
You need access to USB Debugging or a Recovery of course..
thorn obsidian
@cedar pelican "any weird internet tricks to bypass this?"
yes
basic auth
raw.githubusercontent.com/Sharpz7:pw@evil.com
still connects to evil.com, while supplying the "correct" url as an username
@thorn obsidian requires root or just debugging?
@thorn obsidian No clue, I go the recovery route
recovery has the issue of crypto, though
Crypto?
yep, android version 7 (IIRC?) encrypts the data partition using the lock screen pin by default
so you'd need the pin itself to disable the lock screen
which kind of defeats the point
You sure?
I don't think that's forced.
so it was 5, not 7, my bad
Yeah, I definitely haven't seen that.
I don't think that's forced. Very much optional.
i'm... pretty sure it isn't
otherwise, stuff like forceencrypt disabler tweaks etc wouldn't be so prevalent in the modding community
i can ask my colleague, he's an android security expert
¯_(ツ)_/¯
I haven't seen them in the devices I've worked on
cedar pelican
@thorn obsidian Thanks a lot. I knew there was something like that. To fix that, can I just remove all colons from a URL string?
mellow steeple
do anyone know what is the network traffic type of discord text messages?
gentle light
Websockets and REST?
thorn obsidian
@mellow steeple What are you trying to do with Discord messages?
spiral turtle
gg
mellow steeple
scott,
track someone's IP?
kindred wraith
I'm not sure that's possible. My understanding is that discord does not do any kind of peer-to-peer connections. So, if you did find the traffic with wireshark, you'll only be able to see IP of discord servers, afaik. @mellow steeple
You can read about the discord architecture here: https://blog.discordapp.com/how-discord-handles-two-and-half-million-concurrent-voice-users-using-webrtc-ce01c3187429
In relevant part:
Routing all your network traffic through Discord servers also ensures that your IP address is never leaked whether you use text, voice, or video — preventing anyone from finding out your IP address...
mellow steeple
yea...i read it before..
but i saw a vid about tracking ip of discord user when they getting a voice call
so i just needed to try
kindred wraith
If that's true, to me that sounds like it'd be a bug/vulnerability with discord.
Unfortunately, we probably wouldn't be able to help with exploiting that (even if just for testing/poc) -- per rule #5 of the server.
mellow steeple
yea..i understand just needed to know
thorn obsidian
@mellow steeple That's not how Discord works at all
Whatever you read/saw is wrong, as everything is done through the servers.
mellow steeple
probably i have a misunderstanding about what i saw
fathom wadi
Hey guys i had a question i am trying to build an application and i want to have an auth server so when it launches its prompts the user to enter a key if they key is invalid then the app won't run if it is valid the app will run and the keys can only be used one time so people can't share keys if that makes sense any suggestions on where to start
orchid notch
Are you trying to describe what is basically a licensing system
fathom wadi
Yes @orchid notch
cedar pelican
How do I make self-signed certificates that will work on my hosts IP localhost?
I want to make them automatically... I know this isn't ideal, I was hoping to get some help coming up with a solution
cedar pelican
@thorn obsidian How do I go around getting them up with a single command?
I need them to be created automatically with no user involvement
thorn obsidian
Gimme a sec
cedar pelican
btw after hours of trying to get sharpops off of root I did it
thorn obsidian
quick venture
@fathom wadi uuid check + server check
but i think that my grandma can break licensing in python
shadow karma
I'm asking here since it seems this is the better place to do so:
if a web server uses cookies for something like a login system, and these cookies are signed, and the session is stored in the client side, is there anything that prevents someone to copy and use those cookies from a different client? (I know there are ways to prevent cookie stealing from a different page in the browser itself, but I'm talking about manually grabbing the cookie)
I'm trying to figure out what I could share (or not) in a cookie, since sometimes there's information that is private but not strictly sensitive, I'm wondering if I could keep something like that temporally stored in the client's side
cedar pelican
@thorn obsidian Certbot needs to have a valid domain attached
So that's not a option
shadow karma
well, nerdwaller's answer was what I was looking for, among other things: https://www.reddit.com/r/flask/comments/5l2gmf/comment/dbtjjzn
thorn obsidian
@cedar pelican Was under the impression you were using a domain.
cedar pelican
@thorn obsidian I am, but this bit of software cannot. I am going to make docs for this eventually, letting people know that certbot is the best way to do this.
thorn obsidian
The software doesn't need to be the one that sets up the domain. You just have to have a domain and then have the software used on the server..
I imagine the software is using nginx/apache or something?
fathom wadi
@quick venture I honestly don't mind if its possible for someone to be able to break the license do you know where i should start with uuid check + server check ?
cedar pelican
@thorn obsidian No... I actually got https working natively with golang.
The only thing I need is encryption, I can let users setup the bot themselves.
Ill have sharpops setup with nginx and docker-compose, they just have to configure it manually
Cause I can't find any way to make it automatic by default without forcing users to buy a domain
cedar pelican
Can I ask, does a self-signed certificate actually offer that encyption? Im guessing it does but I want to make sure.
quick venture
@fathom wadi well, u need to get hwid of user's machine. you can check md5(md5(hwid)+secret)), and if it correct you can do a request to your web server with user's license and if it's correct send true to client
it's just an example
fathom wadi
What server would you recommend using
orchid notch
@cedar pelican a self signed certificate is not at all different form a regular one except that it has another CA
cedar pelican
Yeah so its all good
thorn obsidian
self-signed isn't good
@cedar pelican ^
How are you going to check if a self-signed cert has been revoked?
cedar pelican
@thorn obsidian The certificate is there to deal with encryption. You only ever connect to your own server, with your own services.
I honestly can't think of a better way to do it.
Again, hopefully I can make it clear in the docs self-signed isnt the way to go, but I'm not going to stop users doing it, as its what I need for my usecase
arctic vale
@cedar pelican https://letsencrypt.org/docs/certificates-for-localhost/ details what you need to do, in essence you need to generate your own certificate, and add it to the clients locally trusted certs. This is the only right way of setting up https for localhost/127.0.0.1.
Why do you need https for localhost? Is it for development purposes or is it just to be "secure"?
cedar pelican
I'm using it for consistency. My app is a daemeon that runs commands for you, a bit like tmux, but you can send commands remotely.
Since my client location changes, and i cant release the private key, using that for this isn't the ideal usecase, but I can definitely try and think about it.
grizzled lake
im not sure how using https on localhost helps though
thorn obsidian
arctic vale
@cedar pelican It sounds that https might not be the right way for the problem since the client changes, and there being no domain. If remote clients forward traffic over SSH you do not need https, but the traffic will still be encrypted OtW.
cedar pelican
SSH might work, but i can't send automated requests like a webhook.
I need to send encrypted "POST like" data. If there is a way to do this with SSH then that's great
But I still feel like this won't reach the automation level i'm aiming for
I'm making sure people know that nginx + certbot is the way to go
But for what I'm using my software for, I control my nginx certbot setup with
arctic vale
Yeah, you can forward http traffic over SSH, so the client uses SSH as a proxy, and then sends the traffic over that.
cedar pelican
But I still need the user to setup the private and public keys, move the private key to all the places you use it.... its just not ideal
I prefer just having tokens
thorn obsidian
@cedar pelican ok so
draw me a graph or something of exactly what the fuck you want happening and encrypted
cedar pelican
@thorn obsidian It's all open source, if you understand golang, thats what it's written in
It's all working, I just want to harden it a bit
I'll draw a graph later
My website currently uses sharpops to deploy, so if you read the circleci config and the .sharpops.ini, you will understand what's going on
thorn obsidian
if you understand golang
lmao my good friend
i wrote my final dissertation in golang, i think i'm ok with it
cedar pelican
Usually people here hate golang, that's why I said it hahaha
Nice to know there's another golang fan in the house
(The code is crap, I know, I'm going to go back and make it better)
thorn obsidian
nah i like golang
it's fun to write, which is pretty important to me
i've got some public projects in go as well iirc
grizzled lake
there are some things i like about it, i just dont think its very fun to program with
spiral turtle
I think that the ssl module is nuts.
tepid venture
So I'm creating some sort of chatting system with websockets, would encrypting the text to the server be a good idea? what steps should i take to making this secure, this is my first time making this type of thing
thorn obsidian
I'd work with things that already exist like Signal if I was setting it up.
tepid venture
@thorn obsidian Signal?
thorn obsidian
@tepid venture https://en.wikipedia.org/wiki/Signal_Messenger
Specifically https://en.wikipedia.org/wiki/Signal_Protocol
There's a C, Java, and JavaScript implementation
tepid venture
Then what about Python? @thorn obsidian
grizzled lake
your best bet is probably to bind to the C implementation using python's CFFI
tepid venture
So I'm reading the Java docs on github, cause I'm more experienced with Java than C
but
IdentityKeyPair identityKeyPair = KeyHelper.generateIdentityKeyPair();
int registrationId = KeyHelper.generateRegistrationId();
List<PreKeyRecord> preKeys = KeyHelper.generatePreKeys(startId, 100);
SignedPreKeyRecord signedPreKey = KeyHelper.generateSignedPreKey(identityKeyPair, 5);```like why do these need to be generated?
and what is done with these
cause I think I'm just gonna use cffi and rather not create my own implementation
grizzled lake
Probably crypto stuff
thorn obsidian
I'm not qualified to answer
tepid venture
Cause that's under Install time
it doesnt give a reason why it needs to generate that stuff
`At install time, a libsignal client needs to generate its identity keys, registration id, and prekeys.
`
grizzled lake
Presumably they are used later in the application?
tepid venture
im not sure cause its not used when creating the session
SessionStore sessionStore = new MySessionStore();
PreKeyStore preKeyStore = new MyPreKeyStore();
SignedPreKeyStore signedPreKeyStore = new MySignedPreKeyStore();
IdentityKeyStore identityStore = new MyIdentityKeyStore();
// Instantiate a SessionBuilder for a remote recipientId + deviceId tuple.
SessionBuilder sessionBuilder = new SessionBuilder(sessionStore, preKeyStore, signedPreKeyStore,
identityStore, recipientId, deviceId);
// Build a session with a PreKey retrieved from the server.
sessionBuilder.process(retrievedPreKey);
SessionCipher sessionCipher = new SessionCipher(sessionStore, recipientId, deviceId);
CiphertextMessage message = sessionCipher.encrypt("Hello world!".getBytes("UTF-8"));
deliver(message.serialize());```unless im blind
oh wait
IdentityKeyPair identityKeyPair = KeyHelper.generateIdentityKeyPair(); maybe this just needs to be initialized to create IdentityKeyStore identityStore = new MyIdentityKeyStore();
thorn obsidian
No clue, that's more a thing to ask in the Github or somewhere else tbh. Not sure if there's any qualified cryptographers here.
tepid venture
Yeah I thought, either way thanks for introducing it though
thorn obsidian
No problem 😄
thorn obsidian
@primal compass What are the questions?
primal compass
Ok so I am a student in network and information security, and I am entering to the final year, and I have graduation project to do.
So my questions are:
-
in my university we don't take python at all, so what is the best source to study python and especially for security?
-
I want my project to be in python, do you any Ideas? It must has security part of it
Do you have*
past starBOT
thorn obsidian
In regards to projects, that really depends on your Python level, which if you're just starting out, I'm not sure what to recommend.
primal compass
I have time to learn, I have like 5 monthes before I start to implement
thorn obsidian
Well, think of an actual problem you face security-wise and tackle it
thorn obsidian
I don't dabble in C++, no
spiral turtle
I did a course of C++ for 2 weeks during the summer
it was very weird for a first time especially doing complex mathematical simulations without doing the maths theory before
😂
safe bear
@primal compass If it's a significant project, you should probably stick with a language you know well
thorn obsidian
safe bear
Aleph security huh
thorn obsidian
``Valve patches recent Steam zero-days (both of them), calls turning away researcher 'a mistake'
- also updates bug bounty program rules to accept LPEs``
https://www.zdnet.com/article/valve-patches-recent-steam-zero-days-calls-turning-away-researcher-a-mistake/
safe bear
You're late on that tidbit Scotty boi
thorn obsidian
I didn't know they patched it
covert briar
I have read in reddit Windows Credential manager is about as secure as a plaintext file. How much of true there is in that sentence?
I wanted to use it to store OAuth tokens
I know git for windows uses it to store github OAuth token
(I can see its entry there)
thorn obsidian
@covert briar Where'd you read this?
thorn obsidian
https://security.stackexchange.com/a/120001 and all of the comments on that answer tend to paint a pretty good picture
true wasp
lime robin
meh
thorn obsidian
CamScanner was found with a Trojan Dropper module hidden within it. Might want to get rid of it... https://www.bleepingcomputer.com/news/security/trojan-dropper-malware-found-in-android-app-with-100m-downloads/
Good chance you or someone you know probably has this installed.
orchid notch
(the chance is actually just aroudn 4 percent by the way)
hollow prawn
@thorn obsidian yikes, thanks for the heads up.
thorn obsidian
@hollow prawn No problem 👍
thorn obsidian
@sly atlas What specifically?
thorn obsidian
I have no clue what it is you want to know/learn
sly atlas
like blackhat hacking
but python language
or ethical hacking
but i want to learn hacking is the main point
@thorn obsidian ??
thorn obsidian
!resources
past starBOT
thorn obsidian
Probably Black Hat Python
Huh, it's not in the resources. https://nostarch.com/blackhatpython
@sly atlas To verify, this is for your own system(s), yes?
sly atlas
yes
thorn obsidian
Okay, neat.
thorn obsidian
Yes. You can get it for $1 if you do Humble Bundle: https://www.humblebundle.com/books/python-programming-no-starch-books
grizzled lake
its literally $1
even if you're a kid on allowance you've probably spent more money on dumber stuff
thin mountain
hello I'm a beginner and I want to learn write a malware in python
upbeat palm
I'm afraid no one gonna help you with that
upbeat palm
Specifically malware writing, I don't think so.
thin mountain
Oh ok
violet notch
Malware is a very broad term try being more specific ie. what kind of malware. Then look at what that does and how it works at that point you should have a idea on what libraries to use etc but like the other person said no one is going to tell exactly how to write that
stark vortex
not sure if it's the proper place for this, but the security implications could be important
thorn obsidian
@stark vortex 100% proper place for this, appreciate you posting it
More of a reason not to use Facebook 😄
gentle heron
looks like more data for them to sell. i see a lot of companies, particularly facebook, hashing data, uploading it, then comparing the hashes to a db of known similar info. it lets them upload 'anonymous' data but still know whats in it
thorn obsidian
"anonymous"
violet notch
why do does the uk call grandmas "nans"
sharp remnant
Nans ? Really ?
thorn obsidian
why did you post that in this channel lmao
thorn obsidian
https://www.theguardian.com/technology/2019/aug/30/hackers-monitoring-implants-iphones-google-says An unprecedented iPhone hacking operation, which attacked “thousands of users a week” until it was disrupted in January, has been revealed by researchers at Google’s external security team.
https://thehackernews.com/2019/08/dds-safe-dental-ransomware-attack.html DDS Safe, an online cloud-based data backup system that hundreds of dental practice offices across the United States are using to safeguard medical records and other information of their patients from ransomware attacks has been hit with ransomware.
stark vortex
Started reading the ios exploit one earlier, quite technical but well explained
thorn obsidian
that last one like
lmao
``Google expanded the scope of its Google Play Security Reward Program (GPSRP) to include all Android apps from the Google Play Store with over 100 million installs.
Security researchers who report vulnerabilities in one of these apps will be able to collect rewards from Google and one from the app's developers if they also run their own bug bounty program on the HackerOne platform.``
Twitter
The phone number associated with the account was compromised due to a security oversight by the mobile provider. This allowed an unauthorized person to compose and send tweets via text message from the phone number. That issue is now resolved.
Seeing lots of SIM swapping attacks recently.
thorn obsidian
https://addons.mozilla.org/en-US/firefox/addon/multi-account-containers/
Great extension to use for keeping groups of sites separated. Whether that's Banking, Shopping, etc.
thorn obsidian
anyone here going to Disobey 2020?
thorn obsidian
Who here uses security add-ons like HTTPS Everywhere, NoScript, or the Multi-Account Containers above?
tough rain
NoScript + uMatrix
gentle light
None of those, although I do use privacy badger
Which, is somewhat security related
tough rain
With uMatrix, I set rules to default to block all the things. Any new site I visit gets a warning page from uMatrix. I then have to explicitly allow html/css/javascript/xhr/whatever before the first network request can proceed. Handy for work which sends phish email tests now and then. Accidentally clicking on one sends me to a large warning page instead of immediately failing the test.
thorn obsidian
@tough rain That's pretty smart, actually.
tall haven
What security risks are there in html/css?
I suppose I have heard of some exploits before
tough rain
automatically downloading pages from third party sites
tall haven
And that is bad why?
tough rain
increases the scope of visibility of you
it can also make pages take longer to load. not really a security issue but super annoying
tall haven
I don't know what that means
tough rain
Scope of visibility in this context means how many sites directly know your activities
HTML and CSS can also be used for more than just rendering the webpage. They can be used for fingerprinting who you are.
tall haven
I see
I figured that was what you meant
I'm not too concerned about that
But you make a point on slower loading
tough rain
You're not concerned about adversaries knowing your browsing history?
tall haven
No
I don't have adversaries anyway
I'm not trying to undermine your practices
If you want to take it to that level it's perfectly fine
thorn obsidian
@tall haven HTML/CSS? Not much, really. Fingerprinting mostly. SQL injection, XSS, CSRF are mostly a JavaScript thing
@tough rain - automatically downloading pages from third party sites That's more or less just having 1x1 iframes of a million sites, lol.
For example, I can try to load local files that only work for Firefox, or Chrome, to fingerprint your browser.
Trying to load chrome://branding/content/icon32.png in Firefox obviously works, where in Chrome it doesn't
Pretty sure you can do the same thing with sites that require you to login
So you can then use that to see if they have an account there
But these are all known attacks and trivial to get around.
Also, https://addons.mozilla.org/en-US/firefox/addon/multi-account-containers/ helps for that
tough rain
Indeed? I'll check it out
thorn obsidian
Your containers can't find out anything from the other containers
So if you have a Banking container, anything outside of that container has no clue you're logged into it
😄
@tough rain Now, that's only for preventing sites from seeing if you're logged in somewhere
I'm not sure how you'd prevent someone from checking to see if you have Firefox/Chrome/etc by checking something like chrome://branding/content/icon32.png other than uMatrix/uBlock Origin on advanced, and making sure it can't connect to 3rd party.
grizzled lake
@thorn obsidian i use multi-account and ublock matrix
i actually am considering switching back to ublock origin, which allows blacklisting of individual files
so i could blacklist the tracking script from the opaque cloudflare cdn domain, but whitelist the functionality script
tall haven
Wasn't aware that could be done
Actually useful cause I need to do that for Google captcha
thorn obsidian
@thorn obsidian fun tidbit, the chrome:// uri scheme has been in use by Firefox for longer than Google Chrome has been around
@thorn obsidian I'm well aware. I've been using Firefox since before 2.0 - which introduced tabs! 😄
tabs were amazing when they first came out and gained popularity
Not really a chat for #cybersecurity
cedar pelican
Is it just me that honestly doesn't give a damn whether these companies know what I do?
The line for me is my own thoughts, and what I say to friends in real life
Anything is is a bit meh
grizzled lake
there are a few specific concerns
data compromise -> identity theft or other personal security risks
same w/ a malicious government making malicious use of collected data
those are outside risks to be fair, but that doesnt mean we shouldnt care about them
there are also more mundane things, like your medical history being used against you
private enterprises making decisions against you without your knowledge
its not 100% about "don't collect it", it's about "give me control over where, when, how much, and who it's shared with"
that's all very complicated so the default is "id rather not have you collect it in the first place"
if none of that bothers you then, your choice
tough rain
I don't think it's just you who thinks that way. I do think most people who think that way don't really understand how companies knowing what you do can then influence you to benefit the company at your detriment.
It's also not just about you. What you do provides information about the people around you too.
thorn obsidian
@cedar pelican That type of thinking goes into the mindset of "I have nothing to hide so I have nothing to fear." Which can very easily be defeated with asking you for your bank account information, a copy of your ID, your address, phone number(s), asking you as to why you have a lock on your door or why you'd ever whisper anything to anyone. Because humans do have things to hide, and that's perfectly acceptable. Not everything needs to be aired to everyone, and we should be able to pick/choose who shares what, and if they even have that information to begin with.
tall haven
It's more "I have some thing to hide but the consequences of it being revealed are not big enough to warrant doing anything about it"
thorn obsidian
You say that, but it's different for each person. Threat modeling helps a lot when it comes down to what you feel you should do to protect what you hold important.
tall haven
I was just going off what you said
Your point was that people do have something to hide
so I adjust your quote
Or is your point that people do have something to hide and they are falsely thinking they don't have anything to fear?
thorn obsidian
Some people, yes. But that's a case-by-case basis. Some people may genuinely believe ( falsely ) they have nothing to fear/worry about.
thorn obsidian
I'm just going to come visit you every Saturday uninvited, and walk through your house, looking in every cupboard and drawer. I mean, you have nothing to hide, so what's the problem?
tall haven
If I cant sense your presence and nothing is disturbed then whatever
thorn obsidian
🤔
tall haven
I mean like if I'm not home while you do it 😄
thorn obsidian
I'm then going to tell my lads over at the billboard company what you have in your drawers, so they know what billboards to set up next to your house
not only that, I'm going to keep track of what you have and what might be missing - things that you need that you don't even know yet
tall haven
I've wondered if the concept of targeted advertising is inherently bad
Like if we ignore any sense of violation of privacy that makes the practice feasible
Isn't it actually nicer to see ads for things you actually may be interested in, assuming you were going to see some ad regardless?
thorn obsidian
I've yet to see an ad I cared about, actually.
Yet again, I go out of my way to block ads/scripts, so not exactly the right person to ask for that. 😄
tall haven
Only ads I see are those with pretty ladies on them ( ͡º ͜ʖ ͡º)
But on a serious note, I share that sentiment I suppose
Like for traditional ads
But you have things like recommended items on Amazon which could be considered targeted 🤷 Those I've found useful
thorn obsidian
Those were actually pretty hilarious recently. I got a new phone and all "repeat" buys were screen protectors. Every. Single. One.
I've never bought a screen protector in my life, so that was weird and pretty funny. Since they can't be that good if they're repeat buys 😄
tall haven
That's something I admittedly miss when I ditched google search
It was nice it knew what subjects I searched for often
and prioritised results related to those subjects
at the cost of, you know..., tracking me
thorn obsidian
I ditched Google search a long time ago. Was a user of Scroogle when that was a thing 😄
tall haven
Generally I think most would agree those kinds of personalised search results are good
except on the off chance you're searching for something that's not related to your common searches
thorn obsidian
I think most don't understand what they're giving up when it comes to any level of personalized search. That's not even getting into Google datamining your e-mails or anything else they do.
tall haven
Yes you are right
luxury vs privacy
That's really the whole thing with the origin of this discussion
It applies to that too
Being concerned with your privacy unfortunately is the side of the coin which requires effort
thorn obsidian
A local or remote attacker can execute programs with root privileges. - time to patch! 😄
https://github.com/rapid7/metasploit-framework/pull/12283 BlueKeep exploit in Metasploit
thorn obsidian
I've been investigating the Exim CVE, made a poc for it (default Ubuntu 19.04)
it's quite interesting
white glen
Are there any encryption methoods form the 70s or 80s that are still safe today?
white glen
just curious if anything stood the test of time
thorn obsidian
Not really, no
orchid notch
Right
There is for example RSA
Which is with sufficient key length still safe
So yes there surely are
@white glen
Well technically we don't use it exactly like in the originally proposed paper anymore
There are a few constrains on some input values and we use a new prime test
But apart from that it has been pretty much the same since then
Or the first idea of the diffie Hellman key exchange, also from the 70s
And I'm pretty sure you'll find both of those still widely used
(also fun fact about RSA, whilst being officially discovered in 76 it was actually first discovered and written down by a GCHQ mathematician as a sort of side note on cryptography in 73 (which was before diffie and Hellman even published the paper which led to the hunt for RSA) but the GCHQ wouldn't publish it)
white glen
@orchid notch thank you for the insight.
If they just sacrificed computing time for convience it sounds like it could sitll hold up today.
orchid notch
Oh RSA and diffie hellman key exchange still hold up today and they will continue to until larger scale quantum computers become viable
white glen
neat. I don't know much about security. But I know that 256bits of something is pretty standard, did they use something like 64 bit in the past?
orchid notch
That's for symmetric keys
RSA secure key length is at the minimum 2048 and ideally 4096 bits nowdays
white glen
what was is back then?
white glen
🙂
orchid notch
Probably somewhere around 200 digits decimal
orchid notch
That's around 665 bits so the original key length was proooobavly 512 or 1024
All of cryptography just uses numbers lol
white glen
Yeah I don't knwo much about it lol
orchid notch
It's all just math
orchid notch
What you were talking about before was probably key lengths for symmetric block cipher algorithms like AES
white glen
Yeah
orchid notch
And while technically 128 bit are fully sufficient as key people tend to us 256 because the number is bigger I guess?
As long as your attacker doesn't have what's probably more then all computers in the world and he actually has to brute force your key and you used a secure mode of operation 128 bit are sufficient
white glen
have any crypto theories been proven bad or wrong years later?
orchid notch
In the 70s we used to use DES instead of AES but that was by design horribly bad (also guess who had influence on the design and changed the original cipher to put a cap on the maximum key length, yes it is the NSA) and broken a few years later, even at its creation it was really just a matter of time
The max key length was 56 bigger, not because people could decide it but because the NSA influenced the maths behind it so it wasn't possible to make it bigger
white glen
Is AES getting "cracked" a mtter of computer becoming powerfull enough to render it useless or is it finding a loophole?
orchid notch
At the moment it's a matter of pretty pretty big computing power
People have found ways to attack AES easily if the encryptor does a few things wrong and the one or other way to reduce the amount of required operations by a few powers of 2 which considering its in the 2^100s isn't too bad
So for example while AES 128 bit should take 2^128 operations you can, even with perfectly implemented and performed AES still get there with "just" 2^126
white glen
I see
orchid notch
I mean if somebody actually found a way to break AES in general all of TLS, SSH, disk encryption and what not would be fucked over night
white glen
so much relies on AES
quality over quantity I guess
I gotta go to sleep but good talking to you @orchid notch
thorn obsidian
@orchid notch People use AES-256 as it's withstood the test of time and is used to store TOP SECRET information for the U.S. government. Which, is a pretty good indicator it works. Not because the number is bigger I guess?.
@white glen AES is fairly strong compared to, say, asymetric crypto. Considering quantum computing will destroy the latter.
upbeat palm
Even 256 GCM is used is TLS, considering it's security we are not going to get any new any sooner. It's more or less depends on the implementation, I assume?
thorn obsidian
I can't go into the detailed specs of AES, as I'm not a cryptographer. But if you're curious about that kind of stuff for asymetric crypto, this: https://safecurves.cr.yp.to/ exists
upbeat palm
Asymmetric? Isn't AES is symmetric?
thorn obsidian
Yes, which is why I differentiated the two.
thorn obsidian
@orchid notch I wouldn't say that anything to do with pre-80's is still used today. Sure, Diffie-Helman existed back then ( https://en.wikipedia.org/wiki/Diffie-Helman ), but not like how it is today. Same thing with cryptography 100 years ago. It existed, sure, but it's changed dramatically 😄
@upbeat palm I've played about with security for 15~ years now 😄
upbeat palm
How old are you? I mean as per your profile pic you look around ~25
warm abyss
@thorn obsidian asymetric crypto will not be destroyed by quantum computing, RSA's not the only asymetric crypto algorithm 😄
thorn obsidian
@warm abyss Citation?
Surely you know about Shor's algo?
( https://en.wikipedia.org/wiki/Shor's_algorithm for the uninitiated )
warm abyss
Well i didn't 😮
thorn obsidian
Quantum computing will 100% destroy asymetric crypto
warm abyss
Leave my ECC alone 😢
thorn obsidian
It's the entire reason Google worked/is working on New Hope - https://security.googleblog.com/2016/07/experimenting-with-post-quantum.html
Yes, that's a Star Wars reference. Buncha nerds these Google fellas... 😄
warm abyss
What about the Quantum Computing's phantom menace over asym crypto ? 👻
thorn obsidian
Specifically, if large quantum computers can be built then they may be able to break the asymmetric cryptographic primitives that are currently used in TLS, the security protocol behind HTTPS.
@upbeat palm My profile picture isn't me, but that's off-topic.
upbeat palm
Oh, no problem. But do share the resources and feel free to tag me along, so I won't miss good conversations?
thorn obsidian
Sure! 😄
orchid notch
@thorn obsidian there is simply no point in AES 256 over 128, both are impossible to attack as of today, that's my point
And on the diffie Hellman part, he just asked, is there crypto from the 70s which is still around today and yes it surely is
@warm abyss interestingly enough even though ECC with lower bit lengths than RSA is significantly harder to attack than RSA it is also easier to break with quantum computers than equally secure RSA algorithms
thorn obsidian
@orchid notch I beg to differ. AES-256 should be the standard if you're using AES at all.
orchid notch
I mean you can have the unbreakable even more unbreakable sure
But in the end the degree of security as of today is for both astronomically high
thorn obsidian
@orchid notch To your knowledge, is there any level of a significant speed decrease between using AES-128 and AES-256?
orchid notch
There should be one by a few nano seconds during the initial key schedule
thorn obsidian
I don't see how that's any level of significant. There also shouldn't be any level of a significant speed decrease between the two, considering hardware support within the processors on anything modern 😄
Which again, is more of a reason to use AES-256 - as you don't get a speed decrease.
@upbeat palm ^ If you're interested in AES
upbeat palm
I am. Thanks for the tag :P
thorn obsidian
Actually, looks like hardware support goes back 11 years https://en.wikipedia.org/wiki/AES_instruction_set#x86_architecture_processors
orchid notch
I mean you have to generate almost one third of round keys more than 128 compared to 256
And of course it does
AES's design criteria was to be easily implementable in hardware
thorn obsidian
Have you ran benchmarks? If so, what hardware are you running these benchmarks on and are you running the benchmarks on individual files or full disks?
orchid notch
Every idiot who knows how to make an XOR Gate Out of Transistors could in fact build an AES ECB circuit
Full disk encryption is an eeeentirely different topic
Full disk encryption is incredibly easier to get wrong than file encryption
thorn obsidian
Well I'd hope you aren't designing your own full disk encryption schemes, software, or anything to do with it. Especially considering LUKS/VeraCrypt exist and are recommended by anyone in security.
You'll also note that these programs don't use AES-128 😄
orchid notch
In full disk encryption it is a lot more relevant how you do the actual encryption
Because there common things like CBC are gonna fuck you up, hard
thorn obsidian
Well, let's get into that then.
VeraCrypt supports parallelized[20]:63 encryption for multi-core systems and, under Microsoft Windows, pipelined read and write operations (a form of asynchronous processing)[20]:63 to reduce the performance hit of encryption and decryption. On processors supporting the AES-NI instruction set, VeraCrypt supports hardware-accelerated AES to further improve performance.[20]:64 On 64-bit CPUs VeraCrypt uses optimized assembly implementation of Twofish and Camellia.[9]
I don't really see how VeraCrypt/LUKS are doing any level of encryption wrong, though.
I'm also not qualified to audit either, but believe in VeraCrypt considering it was audited 😄
Either way, circling back to the initial point, AES-256 should be used every single time you use AES, and using anything else ( including AES-128 ) should be avoided and not recommend on anything made in the last 10 years.
orchid notch
Consider for example this article
Even though AES XTS is beloved by many it still has many flaws
Disk encryption is simply not a solved problem yet
thorn obsidian
Yes it is
thorn obsidian
Yes, it is.
orchid notch
Tell me how you'd authenticate your plaintext then please
Are you gonna store an hmac for every sector alongside the sector?
thorn obsidian
You're speaking to someone who has the system they're on right now full-disk encrypted.
orchid notch
Authenticated ciphers?
thorn obsidian
I don't know where you're going with this. How do you have unauthenticated full disk encryption?
If I type in the password to unlock my disk, that, by definition, is authenticated.
orchid notch
Yeah that's how you view authentication but it means something different in cryptography
thorn obsidian
Authentication - A security measure designed to protect a communications system against acceptance of a fraudulent transmission or simulation by establishing the validity of a transmission, message, or originator.
orchid notch
Yes but you cannot do authenticated full disk encryption without loosing a few nice things you'd love to have
In full disk encryption you usually want two things
A) ofc the authenticated part and B) you don't want to have to decrypt the entire thing for reading a single sector
thorn obsidian
How would you fix this, then?
you don't want to have to decrypt the entire thing for reading a single sector - how would you know where the information you wanted was if you didn't decrypt the entire thing?
orchid notch
So people came up with sector based encryption based upon tweakable ciphers (you can read more about that in the AES XTS article from above, if you want to question it's credibility, it was recommended to me from one of the people at work who contribute to the built-in cryptography of open BSD)
And that would again be a topic for itself which goes into file system design
Now where we encrypt each sector we however have the problem that once we decrypt them we can't really know wether that's actually the correct clear text as methods like XTS aren't authenticated
And with authenticated I mean, provide a way to verify that cipher text c actually got decrypted properly with key k
And yes Vera crypt does use XTS
thorn obsidian
I'm curious as to why you believe there's no way to verify the text was decrypted
orchid notch
Oh there is a way
Store a hash of everies sector clear text next to the sector but that's a little space consuming so we don't do that
Again, read the article from above about XTS and then you might understand that while it's the best for its job it is not perfect
thorn obsidian
Are you aware of LUKS2?
LUKS2 is a new version of header format that allows additional extensions like different PBKDF algorithm or authenticated encryption.
orchid notch
XTS was even publicly criticised by the Nist
For what I just said and othe reasons
And if you don't know the Nist
They're the ones responsible for AES so they might know what they're talking about
thorn obsidian
I'm aware of the National Institute of Standards and Technology.
orchid notch
Wonderful
thorn obsidian
Though, you didn't answer the LUKS2 question. Are you aware of it?
orchid notch
Well we're still not done with the talk about XTS and that it's while the best at the moment not a perfect solution for what it's used for
As I proved with now two sources, one actually official and the other one recommended by one of the people who actually build cryptography implementations
orchid notch
My original statement was that full disk encryption is a not perfectly solved problem, XTS is used by at least one (I'll check up luks soon) of the implementations you tried to bring up as counter arguements and XTS is as I just proved before while the best not a perfect solution
thorn obsidian
Either way, do you use full-disk encrytion?
orchid notch
And for LUKS, it does have authenticated encryption but it uses exactly the mechanism of taking some additional space on the disk which arises a few problems like
If a luks header should be damaged the data gets, depending on what's damaged irrecoverable, this might be especially funny for you on an SSD
It takes up disk space, so if I have a volume limited to X GB which is filled to the brim I can't make a LUKS encrypted FS out of it
Luks does actually leak a bit of usage meta data because of it's mechanism for key slots
So while it does some things better than Vera crypt it has to do a trade off for that, just like Vera crypt does for using XTS, ideally we would want to have a solution which doesn't have a trade off into either direction
And why exactly would that be relevant?
thorn obsidian
I think that's fairly relevant. If you don't use FDE, I'm curious as to why.
If a luks header should be damaged the data gets, depending on what's damaged irrecoverable
Yes, this is well known.
Luks does actually leak a bit of usage meta data because of it's mechanism for key slots
LUKS leaks the fact that it's LUKS based off just the header.
Again, I'm confused. Have you used or are you currently using FDE? Because a lot of what you just brought up isn't new.
It seems to me that you believe there is no verification at all and things may or may not be encrypted.
FDE is also an option upon installation of most Linux distros, so it's normally enabled upon install. Setting it up after the fact isn't exactly straightforward for beginners or anyone who is worried about borking their install.
So I think that before arguing the nuances of XTS this and CBC that, I think it'd be more beneficial to actually use FDE and point out the issues/problems you can see from well established software like VeraCrypt/LUKS on a day-to-day basis.
Because we can argue theoretical all day and it'll get us nowhere.