K. Reid Wightman :verified: 🌻 :donor:<p>In addition to duplicate CVEs, 'the industry' also has some amusing CVE mis-assignment.</p><p>Recently CVE-2017-9833 and CVE-2021-33558 were brought up as attack vectors in a campaign against liquified natural gas victims.</p><p>Both CVEs, it turns out, are mis-attributed to the open source Boa webserver. I am pretty darned confident that neither CVE actually applies to the Boa codebase. And I'm not the only one to notice (see all the way at the end, some actually famous infosec friend noticed one of them). </p><p>Let's take a look:</p><p>CVE-2017-9833 is stated to be a vulnerability in Boa webserver 0.94.14rc21. The vulnerability description ( <a href="https://nvd.nist.gov/vuln/detail/CVE-2017-9833" rel="nofollow noopener" target="_blank"><span class="invisible">https://</span><span class="ellipsis">nvd.nist.gov/vuln/detail/CVE-2</span><span class="invisible">017-9833</span></a> ) is very odd though: apparently it is a directory traversal in a specific CGI handler called /cgi-bin/wapopen . The variable itself is called 'FILECAMERA'. This appears to be a vulnerability in a specific IP camera, not Boa itself. To wit, you can retrieve the original source package for Boa 0.94.14rc21 and search all of the files for the variable 'FILECAMERA' and there is no mention of this variable. Of course it would be very odd for a generic web server to include any cgi handlers — sure you’d expect a generic mechanism to bolt on your own CGI applications but unless there is some ‘test’ application for demonstrating CGI, we wouldn’t expect such a vulnerability to be in the web server itself. A quick check of exposed webservers for the /cgi-bin/wapopen URL shows that they 404 (as we’d expect).</p><p>CVE-2021-33558 is likewise very odd. It is stated to be a vulnerability in Boa webserver 0.94.13. The minimal writeup and public proof of concept just indicate that a few HTML and JS files are exposed without authentication. Again, these are files that are on some specific device firmware. Boa itself does not include a ‘backup.html’ or a ‘preview.html’ or ‘js/log.js’, or any of the files included in the advisory. This is a mistake in one particular device, not in Boa itself. And again if we try to load some of these files on generic Boa webservers, we get a 404.</p><p>Fixing all of this in the public literature is likely an impossible task since we are not the original researchers? First we have to identify exactly what devices the researchers were looking at, and then get those researchers to publish new advisories I guess. In the case of CVE-2017-9833, the researcher has no public point of contact, which is unfortunate. In the case of CVE-2021-33558, our friend attrition (does anyone know if he’s on some mastodon instance?) already tried to get some confirmation because they spotted the same thing we did. There doesn’t seem to be any update though ( <a href="https://github.com/mdanzaruddin/CVE-2021-33558./issues/1" rel="nofollow noopener" target="_blank"><span class="invisible">https://</span><span class="ellipsis">github.com/mdanzaruddin/CVE-20</span><span class="invisible">21-33558./issues/1</span></a> ).</p><p>The bigger picture? If we toss a firmware with Boa into a vulnerability scanning tool, or even run a network scanning tool against a device with Boa, it’s probably going to flag the web server as vulnerable even though it isn’t. And that’s almost not the scanning tool’s fault: it’s really a problem with CVE mis-assignment.</p><p>If you made it this far, thanks for listening to my rant. ⭐</p><p><a href="https://infosec.exchange/tags/vulnerabilities" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>vulnerabilities</span></a> <a href="https://infosec.exchange/tags/cves" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>cves</span></a> <a href="https://infosec.exchange/tags/threatintel" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>threatintel</span></a> <a href="https://infosec.exchange/tags/sbom" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>sbom</span></a></p>