Annotation of www/security.html, Revision 1.456
1.441 bentley 1: <!doctype html>
2: <html lang=en>
3: <meta charset=utf-8>
4:
1.430 tj 5: <title>OpenBSD: Security</title>
1.425 deraadt 6: <meta name="viewport" content="width=device-width, initial-scale=1">
7: <link rel="stylesheet" type="text/css" href="openbsd.css">
1.432 tb 8: <link rel="canonical" href="https://www.openbsd.org/security.html">
1.1 deraadt 9:
1.441 bentley 10: <style>
11: h3 {
12: color: var(--red);
13: }
14: </style>
1.428 tb 15:
1.441 bentley 16: <h2 id=OpenBSD>
1.425 deraadt 17: <a href="index.html">
1.441 bentley 18: <i>Open</i><b>BSD</b></a>
19: Security
1.427 tb 20: </h2>
1.441 bentley 21:
1.294 david 22: <hr>
1.441 bentley 23:
1.429 tj 24: <p>
1.441 bentley 25: For security advisories for specific releases, click below:
1.1 deraadt 26:
1.294 david 27: <p>
1.406 deraadt 28:
1.444 schwarze 29: <a href="errata20.html">2.0</a>,
1.418 tedu 30: <a href="errata21.html">2.1</a>,
31: <a href="errata22.html">2.2</a>,
32: <a href="errata23.html">2.3</a>,
33: <a href="errata24.html">2.4</a>,
34: <a href="errata25.html">2.5</a>,
35: <a href="errata26.html">2.6</a>,
36: <a href="errata27.html">2.7</a>,
37: <a href="errata28.html">2.8</a>,
38: <a href="errata29.html">2.9</a>,
39: <a href="errata30.html">3.0</a>,
40: <a href="errata31.html">3.1</a>,
41: <a href="errata32.html">3.2</a>,
42: <a href="errata33.html">3.3</a>,
43: <a href="errata34.html">3.4</a>,
44: <a href="errata35.html">3.5</a>,
1.444 schwarze 45: <br>
1.418 tedu 46: <a href="errata36.html">3.6</a>,
1.420 schwarze 47: <a href="errata37.html">3.7</a>,
1.418 tedu 48: <a href="errata38.html">3.8</a>,
49: <a href="errata39.html">3.9</a>,
50: <a href="errata40.html">4.0</a>,
51: <a href="errata41.html">4.1</a>,
52: <a href="errata42.html">4.2</a>,
53: <a href="errata43.html">4.3</a>,
54: <a href="errata44.html">4.4</a>,
55: <a href="errata45.html">4.5</a>,
56: <a href="errata46.html">4.6</a>,
57: <a href="errata47.html">4.7</a>,
58: <a href="errata48.html">4.8</a>,
59: <a href="errata49.html">4.9</a>,
60: <a href="errata50.html">5.0</a>,
61: <a href="errata51.html">5.1</a>,
1.444 schwarze 62: <br>
1.418 tedu 63: <a href="errata52.html">5.2</a>,
64: <a href="errata53.html">5.3</a>,
1.420 schwarze 65: <a href="errata54.html">5.4</a>,
1.419 jsg 66: <a href="errata55.html">5.5</a>,
1.420 schwarze 67: <a href="errata56.html">5.6</a>,
1.423 benno 68: <a href="errata57.html">5.7</a>,
1.431 deraadt 69: <a href="errata58.html">5.8</a>,
70: <a href="errata59.html">5.9</a>,
1.434 tj 71: <a href="errata60.html">6.0</a>,
1.435 deraadt 72: <a href="errata61.html">6.1</a>,
1.436 deraadt 73: <a href="errata62.html">6.2</a>,
1.437 deraadt 74: <a href="errata63.html">6.3</a>,
1.440 deraadt 75: <a href="errata64.html">6.4</a>,
1.443 deraadt 76: <a href="errata65.html">6.5</a>,
1.445 deraadt 77: <a href="errata66.html">6.6</a>,
1.446 deraadt 78: <a href="errata67.html">6.7</a>,
1.448 deraadt 79: <br>
1.447 deraadt 80: <a href="errata68.html">6.8</a>,
1.448 deraadt 81: <a href="errata69.html">6.9</a>,
1.453 deraadt 82: <a href="errata70.html">7.0</a>,
1.454 deraadt 83: <a href="errata71.html">7.1</a>,
1.455 deraadt 84: <a href="errata72.html">7.2</a>,
1.456 ! deraadt 85: <a href="errata73.html">7.3</a>,
! 86: <a href="errata74.html">7.4</a>.
1.406 deraadt 87: <br>
1.56 deraadt 88: <hr>
89:
1.278 deraadt 90: <ul>
1.441 bentley 91: <li><h3 id=goals>Goals</h3>
1.22 deraadt 92:
1.441 bentley 93: <p>
1.14 deraadt 94: OpenBSD believes in strong security. Our aspiration is to be NUMBER
1.22 deraadt 95: ONE in the industry for security (if we are not already there). Our
96: open software development model permits us to take a more
1.442 deraadt 97: uncompromising view towards increased security than most vendors are
1.424 tb 98: able to. We can make changes the vendors would
1.27 deraadt 99: not make. Also, since OpenBSD is exported with <a href=crypto.html>
1.45 deraadt 100: cryptography</a>, we are able to take cryptographic approaches towards
1.441 bentley 101: fixing security problems.
1.18 deraadt 102:
1.441 bentley 103: <li><h3 id=disclosure>Full Disclosure</h3>
1.106 deraadt 104:
1.441 bentley 105: <p>
1.45 deraadt 106: Like many readers of the
1.450 benno 107: <a href="https://marc.info/?l=bugtraq">
1.18 deraadt 108: BUGTRAQ mailing list</a>,
1.106 deraadt 109: we believe in full disclosure of security problems. In the
110: operating system arena, we were probably the first to embrace
111: the concept. Many vendors, even of free software, still try
1.441 bentley 112: to hide issues from their users.
1.106 deraadt 113:
1.441 bentley 114: <p>
1.106 deraadt 115: Security information moves very fast in cracker circles. On the other
116: hand, our experience is that coding and releasing of proper security
1.441 bentley 117: fixes typically requires about an hour of work — very fast fix
1.106 deraadt 118: turnaround is possible. Thus we think that full disclosure helps the
119: people who really care about security.<p>
120:
1.441 bentley 121: <li><h3 id=process>Audit Process</h3>
1.15 deraadt 122:
1.441 bentley 123: <p>
1.12 deraadt 124: Our security auditing team typically has between six and twelve
1.45 deraadt 125: members who continue to search for and fix new security holes. We
126: have been auditing since the summer of 1996. The process we follow to
127: increase security is simply a comprehensive file-by-file analysis of
1.106 deraadt 128: every critical software component. We are not so much looking for
129: security holes, as we are looking for basic software bugs, and if
1.138 deraadt 130: years later someone discovers the problem used to be a security
1.106 deraadt 131: issue, and we fixed it because it was just a bug, well, all the
132: better. Flaws have been found in just about every area of the system.
133: Entire new classes of security problems have been found during our
134: audit, and often source code which had been audited earlier needs
135: re-auditing with these new flaws in mind. Code often gets audited
136: multiple times, and by multiple people with different auditing
1.441 bentley 137: skills.
1.12 deraadt 138:
1.441 bentley 139: <p>
1.94 deraadt 140: Some members of our security auditing team worked for Secure Networks,
141: the company that made the industry's premier network security scanning
142: software package Ballista (Secure Networks got purchased by Network
143: Associates, Ballista got renamed to Cybercop Scanner, and well...)
144: That company did a lot of security research, and thus fit in well
1.106 deraadt 145: with the OpenBSD stance. OpenBSD passed Ballista's tests with flying
1.441 bentley 146: colours since day 1.
1.31 deraadt 147:
1.441 bentley 148: <p>
1.34 deraadt 149: Another facet of our security auditing process is its proactiveness.
1.45 deraadt 150: In most cases we have found that the determination of exploitability
151: is not an issue. During our ongoing auditing process we find many
152: bugs, and endeavor to fix them even though exploitability is not
153: proven. We fix the bug, and we move on to find other bugs to fix. We
154: have fixed many simple and obvious careless programming errors in code
155: and only months later discovered that the problems were in fact
156: exploitable. (Or, more likely someone on
1.451 jasper 157: <a href="https://marc.info/?l=bugtraq">BUGTRAQ</a>
1.441 bentley 158: would report that other operating systems were vulnerable to a <q>newly
159: discovered problem</q>, and then it would be discovered that OpenBSD had
1.45 deraadt 160: been fixed in a previous release). In other cases we have been saved
161: from full exploitability of complex step-by-step attacks because we
162: had fixed one of the intermediate steps. An example of where we
1.94 deraadt 163: managed such a success is the lpd advisory that Secure Networks put out.
1.29 deraadt 164:
1.441 bentley 165: <li><h3 id=newtech>New Technologies</h3>
1.278 deraadt 166:
1.441 bentley 167: <p>
1.278 deraadt 168: As we audit source code, we often invent new ways of solving problems.
169: Sometimes these ideas have been used before in some random application
170: written somewhere, but perhaps not taken to the degree that we do.
171:
172: <ul>
173: <li>strlcpy() and strlcat()
174: <li>Memory protection purify
175: <ul>
176: <li>W^X
177: <li>.rodata segment
178: <li>Guard pages
179: <li>Randomized malloc()
180: <li>Randomized mmap()
181: <li>atexit() and stdio protection
182: </ul>
1.295 otto 183: <li>Privilege separation
1.278 deraadt 184: <li>Privilege revocation
185: <li>Chroot jailing
186: <li>New uids
187: <li>ProPolice
1.424 tb 188: <li>... <a href="/innovations.html">and others</a>
1.278 deraadt 189: </ul>
190:
1.441 bentley 191: <li><h3 id=reward>The Reward</h3>
1.106 deraadt 192:
1.441 bentley 193: <p>
1.45 deraadt 194: Our proactive auditing process has really paid off. Statements like
1.441 bentley 195: <q>This problem was fixed in OpenBSD about 6 months ago</q> have become
1.45 deraadt 196: commonplace in security forums like
1.452 tj 197: <a href="https://marc.info/?l=bugtraq">BUGTRAQ</a>.
1.35 deraadt 198:
1.441 bentley 199: <p>
1.45 deraadt 200: The most intense part of our security auditing happened immediately
1.441 bentley 201: before the OpenBSD 2.0 release and during the 2.0→2.1 transition,
1.45 deraadt 202: over the last third of 1996 and first half of 1997. Thousands (yes,
203: thousands) of security issues were fixed rapidly over this year-long
204: period; bugs like the standard buffer overflows, protocol
205: implementation weaknesses, information gathering, and filesystem
206: races. Hence most of the security problems that we encountered were
207: fixed before our 2.1 release, and then a far smaller number needed
208: fixing for our 2.2 release. We do not find as many problems anymore,
209: it is simply a case of diminishing returns. Recently the security
210: problems we find and fix tend to be significantly more obscure or
1.441 bentley 211: complicated. Still we will persist for a number of reasons:
1.36 deraadt 212:
1.35 deraadt 213: <ul>
1.45 deraadt 214: <li>Occasionally we find a simple problem we missed earlier. Doh!
1.35 deraadt 215: <li>Security is like an arms race; the best attackers will continue
1.45 deraadt 216: to search for more complicated exploits, so we will too.
217: <li>Finding and fixing subtle flaws in complicated software is
218: a lot of fun.
1.35 deraadt 219: </ul>
1.441 bentley 220:
1.106 deraadt 221: <p>
1.14 deraadt 222: The auditing process is not over yet, and as you can see we continue
1.441 bentley 223: to find and fix new security flaws.
1.12 deraadt 224:
1.441 bentley 225: <li><h3 id=default><q>Secure by Default</q></h3>
1.106 deraadt 226:
1.441 bentley 227: <p>
1.106 deraadt 228: To ensure that novice users of OpenBSD do not need to become security
229: experts overnight (a viewpoint which other vendors seem to have), we
230: ship the operating system in a Secure by Default mode. All non-essential
231: services are disabled. As the user/administrator becomes more familiar
232: with the system, he will discover that he has to enable daemons and other
233: parts of the system. During the process of learning how to enable a new
1.441 bentley 234: service, the novice is more likely to learn of security considerations.
1.106 deraadt 235:
1.441 bentley 236: <p>
1.106 deraadt 237: This is in stark contrast to the increasing number of systems that
238: ship with NFS, mountd, web servers, and various other services enabled
239: by default, creating instantaneous security problems for their users
1.441 bentley 240: within minutes after their first install.
1.106 deraadt 241:
1.441 bentley 242: <li><h3 id=crypto>Cryptography</h3>
1.106 deraadt 243:
1.441 bentley 244: <p>
1.106 deraadt 245: And of course, since the OpenBSD project is based in Canada, it is possible
246: for us to integrate cryptography. For more information, read the page
1.441 bentley 247: outlining <a href=crypto.html>what we have done with cryptography</a>.
1.106 deraadt 248:
1.441 bentley 249: <li><h3 id=advisories>Advisories</h3>
1.106 deraadt 250:
1.441 bentley 251: <p>
1.418 tedu 252: Please refer to the links at the top of this page.
1.106 deraadt 253:
1.441 bentley 254: <li><h3 id=watching>Watching our Changes</h3>
1.106 deraadt 255:
1.441 bentley 256: <p>
1.21 deraadt 257: Since we take a proactive stance with security, we are continually
258: finding and fixing new security problems. Not all of these problems
1.80 espie 259: get widely reported because (as stated earlier) many of them are not
1.45 deraadt 260: confirmed to be exploitable; many simple bugs we fix do turn out to
261: have security consequences we could not predict. We do not have the
1.441 bentley 262: time resources to make these changes available in the above format.
1.21 deraadt 263:
1.441 bentley 264: <p>
1.21 deraadt 265: Thus there are usually minor security fixes in the current source code
266: beyond the previous major OpenBSD release. We make a limited
1.45 deraadt 267: guarantee that these problems are of minimal impact and unproven
1.44 ian 268: exploitability. If we discover that a problem definitely matters for
1.441 bentley 269: security, patches will show up here <strong>VERY</strong> quickly.
1.21 deraadt 270:
1.441 bentley 271: <p>
1.45 deraadt 272: People who are really concerned with security can do a number of
1.441 bentley 273: things:
1.21 deraadt 274:
275: <ul>
276: <li>If you understand security issues, watch our
1.294 david 277: <a href="mail.html">source-changes mailing list</a> and keep an
1.23 deraadt 278: eye out for things which appear security related. Since
1.21 deraadt 279: exploitability is not proven for many of the fixes we make,
1.441 bentley 280: do not expect the relevant commit message to say <q>SECURITY FIX!</q>.
1.21 deraadt 281: If a problem is proven and serious, a patch will be available
282: here very shortly after.
283: <li>Track our current source code tree, and teach yourself how to do a
1.29 deraadt 284: complete system build from time to time (read /usr/src/Makefile
285: carefully). Users can make the assumption that the current
286: source tree always has stronger security than the previous release.
1.45 deraadt 287: However, building your own system from source code is not trivial;
1.424 tb 288: it is over 850MB of source code, and problems do occur as we
1.45 deraadt 289: transition between major releases.
1.115 ericj 290: <li>Install a binary snapshot for your
1.80 espie 291: architecture, which are made available fairly often. For
1.424 tb 292: instance, an amd64 snapshot is typically made available daily.
1.21 deraadt 293: </ul>
294:
1.441 bentley 295: <li><h3 id=reporting>Reporting problems</h3>
296:
1.9 deraadt 297: <p>
1.441 bentley 298: If you find a new security problem, you can mail it to
1.449 job 299: <a href="mailto:security@openbsd.org">security@openbsd.org</a>.
1.7 deraadt 300: <br>
1.5 deraadt 301: If you wish to PGP encode it (but please only do so if privacy is very
1.112 philen 302: urgent, since it is inconvenient) use this <a href="advisories/pgpkey.txt">pgp key</a>.
1.5 deraadt 303:
1.441 bentley 304: <li><h3 id=papers>Further Reading</h3>
305:
1.107 deraadt 306: <p>
1.389 lum 307: Numerous
1.441 bentley 308: <a href="events.html">papers</a> have been written by OpenBSD team members,
1.389 lum 309: many dedicated to security.
1.294 david 310: </ul>