How NOT to Write Software

| Reading Time:

Advertisement

I was checking our email server for spam one day and saw a password reset email for software we have installed on some of our computers and regularly use. So one of the staff forgot the password which happens often in our business. Only one issue… the email had our password in plain text.

After realising this, I immediately wanted to see how the software actually authenticates against the server. As it was a .NET application, I loaded it in ILSpy on a virtual machine specifically for this, FLARE VM.

Note: This post does contain a lot of the code used in the program, however, the name of the software and other information such as database name, server name, passwords, etc. has been redacted. The program being used is not being used to shame the developer, but instead to guide other developers to creating more secure programs (hence the title, “How NOT to Write Software”). Also, I have never developed with .NET, so the terms that I use may not be correct.

So what does the program do when the login button is clicked?


Remote███DataContext remote███DataContext = new Remote███DataContext();

if
(!remote███DataContext.Connection.ConnectionString.Contains("Password"))

{

remote███DataContext.Connection.ConnectionString = frmMain.sConnectionStr;

}

Let’s get some more information about Remote███DataContext()


public Remote███DataContext()
:base(Settings.Default.███ConnectionString, mappingSource)

{

}

What is the connection string? Well, here it is:


"Data Source=███.co.za,1433;InitialCatalog=███;User ID=███"

So the program is connecting directly to the database, not using a web API. But that connection string doesn’t have a password, how can it connect to a database? At this point, I thought there would be some code to add the logged in user’s password and use that to connect to the database. But let’s go back to the first code sample. frmMain.sConnectionStr could give some information into how it works.


public static string sConnectionStr = "";

Blank… Luckily ILSpy makes it easy to check what else reads or modifies a string.

It is modified by two functions (apologies if this is not the correct term, as previously mentioned, I have not developed with .NET before).

One of them is ███.frmMain.frmMain(). Not much going on here




public static string sConnectionStr = "";



public static string passphrase = "███";

Just setting the connection string to a blank string. You would’ve also noticed that I included a passphrase. I’ll come back to that later. So what else is modifying the connection string? ███.frmMain.tmrStart_Tick is doing something…


sConnectionStr = Settings.Default.███ConnectionString + ";Password=███";

There’s the password. And looking at the 75 other functions in the program that read the connection string, I would assume that the password is used for everything. From the unsecure practices in the code, I also assume that the password can do anything on the database including deleting records, which is a major issue for the type of program it is (again, not revealing much information about the program or its use case, but let’s just say one specific company can lose a lot of money if records were deleted in the database).

The developer did try to add a software registration key that is required to use the program. And asks that a new user calls them in order to obtain the registration key. But why do that when you could just create your own key using the code in the program used to verify the key.


sEncrSeed = ███.███Name.Trim();

int
num2 = 0;

for
(int i = 0; i < sEncrSeed.Length; i++)

{

num2 += (i + 7) * 432 * Convert.ToInt16(sEncrSeed[i]) / 11;

}

Random
random = new Random(num2);

int
seed = random.Next(1723, 9690);

Random
random2 = new Random(seed);

int
seed2 = random2.Next(345, 1679);

Random
random3 = new Random(seed2);

int
num3 = random3.Next(100876, 923453);

if
(███.SoftwareRegKey.Trim() != num3.ToString())

{

frmRegister
frmRegister = new frmRegister();

frmRegister.ShowDialog();

if
(███.SoftwareRegKey.Trim() != num3.ToString())

{

MessageBox.Show("Incorrect Registration Key! Program will now Terminate...");

Application.Exit();

Close();

}

}

Yes, the sEncrSeed is just the name of the company using the software. Let’s use NdT3 as an example. The registration key would then be 320524. And looking at the code, there does not appear to be any server-side validation to this.

Earlier, I included a passphrase in a code sample (public static string passphrase = "███"; which was found in ███.frmMain.frmMain(). What is this passphrase used for?

2019-12-01_13-22

Data encryption. But what is being encrypted?

2019-12-01_13-26

So the passphrase is being used to encrypt something that is used in a login form, password change form, and user creation form. That’s right, passwords are being encrypted. This means that if an attacker does get into the database, they have to decrypt the passwords first. Should be hard enough right? Wrong. A copy of the software is available to download from the same server as the database is on. So if an attacker does get the database, they probably also have the program and can easily extract the encryption key.

You are probably now wondering how that password reset is done? A script on the server decrypting the password then emailing it? No.


Remote███DataContext remote███DataContext = new Remote███DataContext();

if
(!remote███DataContext.Connection.ConnectionString.Contains("Password"))

{

remote███DataContext.Connection.ConnectionString = frmMain.sConnectionStr;

}

IQueryable<RDBUser>
source = from u in remote███DataContext.RDBUsers
where u.LoginName.Trim().ToUpper() == tbxUser.Text.Trim().ToUpper() && (long)u.███ID == frmMain.███ID 
select u;

if
(source.Count() == 1)

{

RDBUser rDBUser = source.Single();

if
(rDBUser.Email.Trim().Length < 5 || !rDBUser.Email.Contains('@'))

{

MessageBox.Show("Invalid or Missing Email address!");

Close();

}

else

{

MailNow(rDBUser.Email.Trim(), rDBUser.Password);

}

}

So it gets the password from the database. The code can be easily modified to get another user’s password and send that password to an attacker’s email address. Then MailNow() does something with it…


MailMessage mailMessage = new MailMessage();

mailMessage.To.Add(sEmail);

mailMessage.Subject = "███ Password for user: " + tbxUser.Text.Trim();

mailMessage.From = new MailAddress(frmMain.sStatementEmailAddr);

mailMessage.Body = "Your password is: " + frmMain.DecryptString(sPassw);

string host = "mail.███.co.za";

if
(frmMain.sSmtpServer.Trim().Length > 5)

{

host = frmMain.sSmtpServer;

}

SmtpClient smtpClient = new SmtpClient(host);

smtpClient.DeliveryMethod = SmtpDeliveryMethod.Network;

smtpClient.UseDefaultCredentials = false;

if
(frmMain.sSmtpServer.Trim().Length > 5)

{

smtpClient.Credentials = new NetworkCredential(frmMain.sMailUser, frmMain.sMailPassword);

}

else

{

smtpClient.Credentials = new NetworkCredential("[email protected]███.co.za", "███");

}

try

{

smtpClient.Send(mailMessage);

tbxEmail.Text = sEmail;

pnlResult.Visible = true;

}

catch
(Exception ex)

{

MessageBox.Show("Email delivery failed:\n\r" + ex.Message);

}

Yes, the program, intended to be installed on computers, takes the password it receives (sPassw), connects to the developer’s email server (with the password hard-coded in) and emails the password to the given email address (sEmail).

So please, if you develop software, do not do any of the things showed in the above code samples. Instead, create a web API (and only allow HTTPS connections to this API), use API keys and hash passwords and store these hashes in the database instead of plain text passwords. The program should only have access to data that it really needs so ensure that requests cannot be modified to access other data in the database.


Back to the blog's home page
Advertisement

Comments