Defensive programming: Difference between revisions

Revision as of 09:31, 12 November 2021

Latest revision as of 21:09, 30 July 2025

(29 intermediate revisions by 23 users not shown)

Line 3:

{{howto|date=March 2012}}

'''Defensive programming''' is a form of [[defensive design]] intended to ensuredevelop programs that are capable of detecting potential security abnormalities and make predetermined responses.<ref>{{Citation |last=Boulanger |first=Jean-Louis |title=6 - Technique to Manage Software Safety |date=2016-01-01 |url=https://www.sciencedirect.com/science/article/pii/B9781785481178500064 |work=Certifiable Software Applications 1 |pages=125–156 |editor-last=Boulanger |editor-first=Jean-Louis |publisher=Elsevier |language=en |isbn=978-1-78548-117-8 |access-date=2022-09-02}}</ref> It ensures the continuing function of a piece of [[software]] under unforeseen circumstances. Defensive programming practices are often used where [[high availability]], [[safety]], or [[computer security|security]] is needed.

Defensive programming is an approach to improve software and [[source code]], in terms of:

Line 10:

* Making the software behave in a predictable manner despite unexpected inputs or user actions.

Overly defensive programming, however, may safeguard against errors that will never be encountered, thus incurring run-time and maintenance costs. There is also a risk that code traps prevent too many [[Exception handling|exceptions]], potentially resulting in unnoticed, incorrect results.

== Secure programming ==

{{main|Secure coding}}

Secure programming is the subset of defensive programming concerned with [[computer security]]. Security is the concern, not necessarily safety or availability (the [[software]] may be allowed to fail in certain ways). As with all kinds of defensive programming, avoiding bugs is a primary objective; however, the motivation is not as much to reduce the likelihood of failure in normal operation (as if safety were the concern), but to reduce the [[attack surface]] – the programmer must assume that the software might be misused actively to reveal bugs, and that bugs could be exploited maliciously.

<syntaxhighlight lang="c">int risky_programming(char *input) {

Line 26:

// ...

}</syntaxhighlight>

The function will result in undefined behavior when the input is over 1000 characters. Some novice programmers may not feel that this is a problem, supposing that no user will enter such a long input. This particular bug demonstrates a vulnerability which enables [[buffer overflow]] [[exploit (computer security)|exploit]]s. Here is a solution to this example:

<syntaxhighlight lang="c">int secure_programming(char *input) {

Line 34:

// Copy input without exceeding the length of the destination.

strncpy(str, input, sizeof(str));

// If strlen(input) >= sizeof(str) then strncpy won't null terminate.

Line 62:

}

return "black"; // To be handled as a dead traffic light.

// Warning: This last 'return' statement will be dropped by an optimizing

// compiler if all possible values of 'traffic_light_color' are listed in

// the previous 'switch' statement...

}

</syntaxhighlight>

Line 77 ⟶ 74:

}

assert(0); // Assert that this section is unreachable.

// Warning: This 'assert' function call will be dropped by an optimizing

// compiler if all possible values of 'traffic_light_color' are listed in

// the previous 'switch' statement...

}

</syntaxhighlight>

Line 126 ⟶ 120:

* [[Legacy code]] may not have been designed under a defensive programming initiative, and might therefore be of much lower quality than newly designed source code.

* Legacy code may have been written and tested under conditions which no longer apply. The old quality assurance tests may have no validity any more.

** '''Example 1''': legacy code may have been designed for ASCII input but now the input is [[UTF-8]].

** '''Example 2''': legacy code may have been compiled and tested on 32-bit architectures, but when compiled on 64-bit architectures, new arithmetic problems may occur (e.g., invalid signedness tests, invalid type casts, etc.).

** '''Example 3''': legacy code may have been targeted for offline machines, but becomes vulnerable once network connectivity is added.

Line 132 ⟶ 126:

Notable examples of the legacy problem:

* [[BIND|BIND 9]], presented by Paul Vixie and David Conrad as "BINDv9 is a [[Rewrite (programming)|complete rewrite]]", "Security was a key consideration in design",<ref>{{Cite web|url=http://impressive.net/archives/fogo/[e-mail]|title=fogo archive: Paul Vixie and David Conrad on BINDv9 and Internet Security by Gerald Oskoboiny <[e-mail]>|website=impressive.net|access-date=2018-10-27}}</ref> naming security, robustness, scalability and new protocols as key concerns for rewriting old legacy code.

* [[Microsoft Windows]] suffered from "the" [[Windows Metafile vulnerability]] and other exploits related to the WMF format. Microsoft Security Response Center describes the WMF-features as ''"Around 1990, WMF support was added... This was a different time in the security landscape... were all completely trusted"'',<ref>{{Cite news|url=http://blogs.technet.com/msrc/archive/2006/01/13/417431.aspx|title=Looking at the WMF issue, how did it get there?|work=MSRC|access-date=2018-10-27|language=en-US|archive-url=https://web.archive.org/web/20060324152626/http://blogs.technet.com/msrc/archive/2006/01/13/417431.aspx|archive-date=2006-03-24|url-status=dead}}</ref> not being developed under the security initiatives at Microsoft.

* [[Oracle Corporation|Oracle]] is combating legacy problems, such as old source code written without addressing concerns of [[SQL injection]] and [[privilege escalation]], resulting in many security vulnerabilities which have taken time to fix and also generated incomplete fixes. This has given rise to heavy criticism from security experts such as [[David Litchfield]], [[Alexander Kornbrust]], [[Cesar Cerrudo]].<ref>{{Cite web|url=http://seclists.org/lists/bugtraq/2006/May/0039.html|title=Bugtraq: Oracle, where are the patches???|last=Litchfield|first=David|website=seclists.org|access-date=2018-10-27}}</ref><ref>{{Cite web|url=http://seclists.org/lists/bugtraq/2006/May/0045.html|title=Bugtraq: RE: Oracle, where are the patches???|last=Alexander|first=Kornbrust|website=seclists.org|access-date=2018-10-27}}</ref><ref>{{Cite web|url=http://seclists.org/lists/bugtraq/2006/May/0083.html|title=Bugtraq: Re: [Full-disclosure] RE: Oracle, where are the patches???|last=Cerrudo|first=Cesar|website=seclists.org|access-date=2018-10-27}}</ref> An additional criticism is that default installations (largely a legacy from old versions) are not aligned with their own security recommendations, such as [http://www.oracle.com/technology/deploy/security/database-security/pdf/twp_security_checklist_database.pdf [Oracle Database]] Security Checklist], which is hard to amend as many applications require the less secure legacy settings to function correctly.

=== Canonicalization ===

Line 142 ⟶ 136:

Assume that code constructs that appear to be problem prone (similar to known vulnerabilities, etc.) are bugs and potential security flaws. The basic rule of thumb is: "I'm not aware of all types of [[security exploit]]s. I must protect against those I ''do'' know of and then I must be proactive!".

===Other Tipsways toof Securesecuring Your Codecode===

* One of the most common problems is unchecked use of constant-size or pre-allocated structures for dynamic-size data{{cn|date=December 2023}} such as inputs to the program (the [[buffer overflow]] problem). This is especially common for [[string (computer programming)|string]] data in [[C (programming language)|C]]{{cn|date=December 2023}}. C library functions like <code>gets</code> should never be used since the maximum size of the input buffer is not passed as an argument. C library functions like <code>scanf</code> can be used safely, but require the programmer to take care with the selection of safe format strings, by sanitizing it before using it.

* Encrypt/authenticate all important data transmitted over networks. Do not attempt to implement your own encryption scheme, use a [[Cryptography standards|proven one]] instead. Message checking with [[CRC]]a hash or similar technology will also help secure data sent over a network.

====The 3 Rules of Data Security====

* All [[data]] is important until proven otherwise.

* All data is tainted until proven otherwise.

* All code is insecure until proven otherwise.

====The 3three Rulesrules of Datadata Securitysecurity====

* All [[data]] is important until proven otherwise.

* All data is tainted until proven otherwise.

* All code is insecure until proven otherwise.

** You cannot prove the security of any code in [[userland (computing)|userland]], or, more commonly known as: ''"never trust the client"''.

These three rules about data security describe how to handle any data, internally or externally sourced:

'''All data is important until proven otherwise''' - means that all data must be verified as garbage before being destroyed.

'''All data is tainted until proven otherwise''' - means that all data must be handled in a way that does not expose the rest of the runtime environment without verifying integrity.

'''All code is insecure until proven otherwise''' - while a slight misnomer, does a good job reminding us to never assume our code is secure as bugs or [[Undefinedundefined Behaviorbehavior]] may expose the project or system to attacks such as common [[SQL injection]] attacks.

====More Information====

Line 165 ⟶ 158:

* [[Assertion (computing)|Assertions]] (also called '''assertive programming''')

* Prefer [[Exception handling|exceptions]] to return codes

** Generally speaking, it is preferable{{According to whom|date=December 2023}} to throw exception messages that enforce part of your [[application programming interface|API]] [[Design by contract|contract]] and guide the developer instead of returning error code values that do not point to where the exception occuredoccurred or what the program stack looked liked, Better logging and exception handling will increase robustness and security of your software{{cn|date=December 2023}}, while minimizing developer stress{{cn|date=December 2023}}.

==See also==

* [[Computer security]]

* [[Immunity-aware programming]]

== References ==