1
0
Fork 0
mirror of https://gitlab.redox-os.org/CoffeeCode/redox-ssh.git synced 2025-12-28 22:12:19 +01:00
redox-ssh/crypto/aessafe/index.html

309 lines
No EOL
18 KiB
HTML
Raw Permalink Blame History

This file contains invisible Unicode characters

This file contains invisible Unicode characters that are indistinguishable to humans but may be processed differently by a computer. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<meta name="generator" content="rustdoc">
<meta name="description" content="API documentation for the Rust `aessafe` mod in crate `crypto`.">
<meta name="keywords" content="rust, rustlang, rust-lang, aessafe">
<title>crypto::aessafe - Rust</title>
<link rel="stylesheet" type="text/css" href="../../normalize.css">
<link rel="stylesheet" type="text/css" href="../../rustdoc.css">
<link rel="stylesheet" type="text/css" href="../../main.css">
</head>
<body class="rustdoc mod">
<!--[if lte IE 8]>
<div class="warning">
This old browser is unsupported and will most likely display funky
things.
</div>
<![endif]-->
<nav class="sidebar">
<p class='location'>Module aessafe</p><div class="block items"><ul><li><a href="#structs">Structs</a></li></ul></div><p class='location'><a href='../index.html'>crypto</a></p><script>window.sidebarCurrent = {name: 'aessafe', ty: 'mod', relpath: '../'};</script><script defer src="../sidebar-items.js"></script>
</nav>
<nav class="sub">
<form class="search-form js-only">
<div class="search-container">
<input class="search-input" name="search"
autocomplete="off"
placeholder="Click or press S to search, ? for more options…"
type="search">
</div>
</form>
</nav>
<section id='main' class="content">
<h1 class='fqn'><span class='in-band'>Module <a href='../index.html'>crypto</a>::<wbr><a class="mod" href=''>aessafe</a></span><span class='out-of-band'><span id='render-detail'>
<a id="toggle-all-docs" href="javascript:void(0)" title="collapse all docs">
[<span class='inner'>&#x2212;</span>]
</a>
</span><a class='srclink' href='../../src/crypto/aessafe.rs.html#7-1231' title='goto source code'>[src]</a></span></h1>
<div class='docblock'><p>The <code>aessafe</code> module implements the AES algorithm completely in software without using any table
lookups or other timing dependant mechanisms. This module actually contains two seperate
implementations - an implementation that works on a single block at a time and a second
implementation that processes 8 blocks in parallel. Some block encryption modes really only work if
you are processing a single blocks (CFB, OFB, and CBC encryption for example) while other modes
are trivially parallelizable (CTR and CBC decryption). Processing more blocks at once allows for
greater efficiency, especially when using wide registers, such as the XMM registers available in
x86 processors.</p>
<h2 id='aes-algorithm' class='section-header'><a href='#aes-algorithm'>AES Algorithm</a></h2>
<p>There are lots of places to go to on the internet for an involved description of how AES works. For
the purposes of this description, it sufficies to say that AES is just a block cipher that takes
a key of 16, 24, or 32 bytes and uses that to either encrypt or decrypt a block of 16 bytes. An
encryption or decryption operation consists of a number of rounds which involve some combination of
the following 4 basic operations:</p>
<ul>
<li>ShiftRows</li>
<li>MixColumns</li>
<li>SubBytes</li>
<li>AddRoundKey</li>
</ul>
<h2 id='timing-problems' class='section-header'><a href='#timing-problems'>Timing problems</a></h2>
<p>Most software implementations of AES use a large set of lookup tables - generally at least the
SubBytes step is implemented via lookup tables; faster implementations generally implement the
MixColumns step this way as well. This is largely a design flaw in the AES implementation as it was
not realized during the NIST standardization process that table lookups can lead to security
problems [1]. The issue is that not all table lookups occur in constant time - an address that was
recently used is looked up much faster than one that hasn&#39;t been used in a while. A careful
adversary can measure the amount of time that each AES operation takes and use that information to
help determine the secret key or plain text information. More specifically, its not table lookups
that lead to these types of timing attacks - the issue is table lookups that use secret information
as part of the address to lookup. A table lookup that is performed the exact same way every time
regardless of the key or plaintext doesn&#39;t leak any information. This implementation uses no data
dependant table lookups.</p>
<h2 id='bit-slicing' class='section-header'><a href='#bit-slicing'>Bit Slicing</a></h2>
<p>Bit Slicing is a technique that is basically a software emulation of hardware implementation
techniques. One of the earliest implementations of this technique was for a DES implementation [4].
In hardware, table lookups do not present the same timing problems as they do in software, however
they present other problems - namely that a 256 byte S-box table takes up a huge amount of space on
a chip. Hardware implementations, thus, tend to avoid table lookups and instead calculate the
contents of the S-Boxes as part of every operation. So, the key to an efficient Bit Sliced software
implementation is to re-arrange all of the bits of data to process into a form that can easily be
applied in much the same way that it would be in hardeware. It is fortunate, that AES was designed
such that these types of hardware implementations could be very efficient - the contents of the
S-boxes are defined by a mathematical formula.</p>
<p>A hardware implementation works on single bits at a time. Unlike adding variables in software,
however, that occur generally one at a time, hardware implementations are extremely parallel and
operate on many, many bits at once. Bit Slicing emulates that by moving all &quot;equivalent&quot; bits into
common registers and then operating on large groups of bits all at once. Calculating the S-box value
for a single bit is extremely expensive, but its much cheaper when you can amortize that cost over
128 bits (as in an XMM register). This implementation follows the same strategy as in [5] and that
is an excellent source for more specific details. However, a short description follows.</p>
<p>The input data is simply a collection of bytes. Each byte is comprised of 8 bits, a low order bit
(bit 0) through a high order bit (bit 7). Bit slicing the input data simply takes all of the low
order bits (bit 0) from the input data, and moves them into a single register (eg: XMM0). Next, all
of them 2nd lowest bits are moved into their own register (eg: XMM1), and so on. After completion,
we&#39;re left with 8 variables, each of which contains an equivalent set of bits. The exact order of
those bits is irrevent for the implementation of the SubBytes step, however, it is very important
for the MixColumns step. Again, see [5] for details. Due to the design of AES, its them possible to
execute the entire AES operation using just bitwise exclusive ors and rotates once we have Bit
Sliced the input data. After the completion of the AES operation, we then un-Bit Slice the data
to give us our output. Clearly, the more bits that we can process at once, the faster this will go -
thus, the version that processes 8 blocks at once is roughly 8 times faster than processing just a
single block at a time.</p>
<p>The ShiftRows step is fairly straight-forward to implement on the Bit Sliced state. The MixColumns
and especially the SubBytes steps are more complicated. This implementation draws heavily on the
formulas from [5], [6], and [7] to implement these steps.</p>
<h2 id='implementation' class='section-header'><a href='#implementation'>Implementation</a></h2>
<p>Both implementations work basically the same way and share pretty much all of their code. The key
is first processed to create all of the round keys where each round key is just a 16 byte chunk of
data that is combined into the AES state by the AddRoundKey step as part of each encryption or
decryption round. Processing the round key can be expensive, so this is done before encryption or
decryption. Before encrypting or decrypting data, the data to be processed by be Bit Sliced into 8
seperate variables where each variable holds equivalent bytes from the state. This Bit Sliced state
is stored as a Bs8State<T>, where T is the type that stores each set of bits. The first
implementation stores these bits in a u32 which permits up to 8 * 32 = 1024 bits of data to be
processed at once. This implementation only processes a single block at a time, so, in reality, only
512 bits are processed at once and the remaining 512 bits of the variables are unused. The 2nd
implementation uses u32x4s - vectors of 4 u32s. Thus, we can process 8 * 128 = 4096 bits at once,
which corresponds exactly to 8 blocks.</p>
<p>The Bs8State struct implements the AesOps trait, which contains methods for each of the 4 main steps
of the AES algorithm. The types, T, each implement the AesBitValueOps trait, which containts methods
necessary for processing a collection or bit values and the AesOps trait relies heavily on this
trait to perform its operations.</p>
<p>The Bs4State and Bs2State struct implement operations of various subfields of the full GF(2<sup>8)</sup>
finite field which allows for efficient computation of the AES S-Boxes. See [7] for details.</p>
<h2 id='references' class='section-header'><a href='#references'>References</a></h2>
<p>[1] - &quot;Cache-Collision Timing Attacks Against AES&quot;. Joseph Bonneau and Ilya Mironov.
<a href="http://www.jbonneau.com/doc/BM06-CHES-aes_cache_timing.pdf">http://www.jbonneau.com/doc/BM06-CHES-aes_cache_timing.pdf</a>
[2] - &quot;Software mitigations to hedge AES against cache-based software side channel vulnerabilities&quot;.
Ernie Brickell, et al. <a href="http://eprint.iacr.org/2006/052.pdf">http://eprint.iacr.org/2006/052.pdf</a>.
[3] - &quot;Cache Attacks and Countermeasures: the Case of AES (Extended Version)&quot;.
Dag Arne Osvik, et al. tau.ac.il/~tromer/papers/cache.pdf.
[4] - &quot;A Fast New DES Implementation in Software&quot;. Eli Biham.
<a href="http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.52.5429&amp;rep=rep1&amp;type=pdf">http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.52.5429&amp;rep=rep1&amp;type=pdf</a>.
[5] - &quot;Faster and Timing-Attack Resistant AES-GCM&quot;. Emilia K ̈asper and Peter Schwabe.
<a href="http://www.chesworkshop.org/ches2009/presentations/01_Session_1/CHES2009_ekasper.pdf">http://www.chesworkshop.org/ches2009/presentations/01_Session_1/CHES2009_ekasper.pdf</a>.
[6] - &quot;FAST AES DECRYPTION&quot;. Vinit Azad. <a href="http://webcache.googleusercontent.com/">http://webcache.googleusercontent.com/</a>
search?q=cache:ld_f8pSgURcJ:csusdspace.calstate.edu/bitstream/handle/10211.9/1224/
Vinit_Azad_MS_Report.doc%3Fsequence%3D2+&amp;cd=4&amp;hl=en&amp;ct=clnk&amp;gl=us&amp;client=ubuntu.
[7] - &quot;A Very Compact Rijndael S-box&quot;. D. Canright.
<a href="http://www.dtic.mil/cgi-bin/GetTRDoc?AD=ADA434781">http://www.dtic.mil/cgi-bin/GetTRDoc?AD=ADA434781</a>.</p>
</div><h2 id='structs' class='section-header'><a href="#structs">Structs</a></h2>
<table>
<tr class=' module-item'>
<td><a class="struct" href="struct.AesSafe128Decryptor.html"
title='struct crypto::aessafe::AesSafe128Decryptor'>AesSafe128Decryptor</a></td>
<td class='docblock-short'>
</td>
</tr>
<tr class=' module-item'>
<td><a class="struct" href="struct.AesSafe128DecryptorX8.html"
title='struct crypto::aessafe::AesSafe128DecryptorX8'>AesSafe128DecryptorX8</a></td>
<td class='docblock-short'>
</td>
</tr>
<tr class=' module-item'>
<td><a class="struct" href="struct.AesSafe128Encryptor.html"
title='struct crypto::aessafe::AesSafe128Encryptor'>AesSafe128Encryptor</a></td>
<td class='docblock-short'>
</td>
</tr>
<tr class=' module-item'>
<td><a class="struct" href="struct.AesSafe128EncryptorX8.html"
title='struct crypto::aessafe::AesSafe128EncryptorX8'>AesSafe128EncryptorX8</a></td>
<td class='docblock-short'>
</td>
</tr>
<tr class=' module-item'>
<td><a class="struct" href="struct.AesSafe192Decryptor.html"
title='struct crypto::aessafe::AesSafe192Decryptor'>AesSafe192Decryptor</a></td>
<td class='docblock-short'>
</td>
</tr>
<tr class=' module-item'>
<td><a class="struct" href="struct.AesSafe192DecryptorX8.html"
title='struct crypto::aessafe::AesSafe192DecryptorX8'>AesSafe192DecryptorX8</a></td>
<td class='docblock-short'>
</td>
</tr>
<tr class=' module-item'>
<td><a class="struct" href="struct.AesSafe192Encryptor.html"
title='struct crypto::aessafe::AesSafe192Encryptor'>AesSafe192Encryptor</a></td>
<td class='docblock-short'>
</td>
</tr>
<tr class=' module-item'>
<td><a class="struct" href="struct.AesSafe192EncryptorX8.html"
title='struct crypto::aessafe::AesSafe192EncryptorX8'>AesSafe192EncryptorX8</a></td>
<td class='docblock-short'>
</td>
</tr>
<tr class=' module-item'>
<td><a class="struct" href="struct.AesSafe256Decryptor.html"
title='struct crypto::aessafe::AesSafe256Decryptor'>AesSafe256Decryptor</a></td>
<td class='docblock-short'>
</td>
</tr>
<tr class=' module-item'>
<td><a class="struct" href="struct.AesSafe256DecryptorX8.html"
title='struct crypto::aessafe::AesSafe256DecryptorX8'>AesSafe256DecryptorX8</a></td>
<td class='docblock-short'>
</td>
</tr>
<tr class=' module-item'>
<td><a class="struct" href="struct.AesSafe256Encryptor.html"
title='struct crypto::aessafe::AesSafe256Encryptor'>AesSafe256Encryptor</a></td>
<td class='docblock-short'>
</td>
</tr>
<tr class=' module-item'>
<td><a class="struct" href="struct.AesSafe256EncryptorX8.html"
title='struct crypto::aessafe::AesSafe256EncryptorX8'>AesSafe256EncryptorX8</a></td>
<td class='docblock-short'>
</td>
</tr></table></section>
<section id='search' class="content hidden"></section>
<section class="footer"></section>
<aside id="help" class="hidden">
<div>
<h1 class="hidden">Help</h1>
<div class="shortcuts">
<h2>Keyboard Shortcuts</h2>
<dl>
<dt>?</dt>
<dd>Show this help dialog</dd>
<dt>S</dt>
<dd>Focus the search field</dd>
<dt>&larrb;</dt>
<dd>Move up in search results</dd>
<dt>&rarrb;</dt>
<dd>Move down in search results</dd>
<dt>&#9166;</dt>
<dd>Go to active search result</dd>
<dt>+</dt>
<dd>Collapse/expand all sections</dd>
</dl>
</div>
<div class="infos">
<h2>Search Tricks</h2>
<p>
Prefix searches with a type followed by a colon (e.g.
<code>fn:</code>) to restrict the search to a given type.
</p>
<p>
Accepted types are: <code>fn</code>, <code>mod</code>,
<code>struct</code>, <code>enum</code>,
<code>trait</code>, <code>type</code>, <code>macro</code>,
and <code>const</code>.
</p>
<p>
Search functions by type signature (e.g.
<code>vec -> usize</code> or <code>* -> vec</code>)
</p>
</div>
</div>
</aside>
<script>
window.rootPath = "../../";
window.currentCrate = "crypto";
</script>
<script src="../../main.js"></script>
<script defer src="../../search-index.js"></script>
</body>
</html>