<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Hey, it&#39;s Asim</title>
    <link>https://heyitsas.im/</link>
    <description>Recent content on Hey, it&#39;s Asim</description>
    <generator>Hugo -- gohugo.io</generator>
    <language>en-us</language>
    <copyright>© 2026 Asim Viladi Oglu Manizada</copyright>
    <lastBuildDate>Wed, 27 May 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://heyitsas.im/index.xml" rel="self" type="application/rss+xml" />
    
    <item>
      <title>CIFSwitch: a non-universal Linux local root vulnerability</title>
      <link>https://heyitsas.im/posts/cifswitch/</link>
      <pubDate>Wed, 27 May 2026 00:00:00 +0000</pubDate>
      
      <guid>https://heyitsas.im/posts/cifswitch/</guid>
      <description>&lt;p&gt;&lt;em&gt;TLDR: CIFSwitch (CVE-2026-46243) is a distro-specific Linux LPE found by harnessing LLMs into better multihop knowledge composition. Read on for affected distros, mitigations, and vulnerability details.&lt;/em&gt;&lt;/p&gt;&#xA;&#xA;&lt;h2 class=&#34;relative group&#34;&gt;Background&#xA;    &lt;div id=&#34;background&#34; class=&#34;anchor&#34;&gt;&lt;/div&gt;&#xA;    &#xA;    &lt;span&#xA;        class=&#34;absolute top-0 w-6 transition-opacity opacity-0 -start-6 not-prose group-hover:opacity-100 select-none&#34;&gt;&#xA;        &lt;a class=&#34;text-primary-300 dark:text-neutral-700 !no-underline&#34; href=&#34;#background&#34; aria-label=&#34;Anchor&#34;&gt;#&lt;/a&gt;&#xA;    &lt;/span&gt;&#xA;    &#xA;&lt;/h2&gt;&#xA;&lt;p&gt;In &lt;a href=&#34;https://heyitsas.im/posts/drinking-llms/#future-research-directions&#34;  target=&#34;_blank&#34; rel=&#34;noreferrer&#34;&gt;Getting LLMs Drunk to Find Remote Linux Kernel OOB&#xA;Writes (and More)&lt;/a&gt;,&#xA;I&amp;rsquo;d mentioned how improving LLMs&amp;rsquo; ability to compose existing knowledge is a promising avenue for unlocking&#xA;&amp;ldquo;creative&amp;rdquo; &amp;ndash; or at least non-trivial &amp;ndash; vulnerability findings. Incidentally, among the latest slew of Linux LPEs,&#xA;&lt;strong&gt;CopyFail&lt;/strong&gt; stood out for &amp;ndash; among other things &amp;ndash; exquisitely composing several logic bugs,&#xA;serving as a reminder of the massive potential value of the approach. Unfortunately, training a capable looped&#xA;transformer to improve compositionality was a non-starter, so I started looking for harness-level improvements instead.&lt;/p&gt;</description>
      <media:content xmlns:media="http://search.yahoo.com/mrss/" url="https://heyitsas.im/posts/cifswitch/featured.webp" />
    </item>
    
    <item>
      <title>Getting LLMs Drunk to Find Remote Linux Kernel OOB Writes (and More)</title>
      <link>https://heyitsas.im/posts/drinking-llms/</link>
      <pubDate>Wed, 29 Apr 2026 00:00:00 +0000</pubDate>
      
      <guid>https://heyitsas.im/posts/drinking-llms/</guid>
      <description>&lt;p&gt;&lt;em&gt;TLDR: the grossly overengineered, self-orchestrating team of vulnerability-hunting agents detailed below has discovered 20+ CVEs over the past few months, including &lt;a href=&#34;https://nvd.nist.gov/vuln/detail/CVE-2026-31432&#34;  target=&#34;_blank&#34; rel=&#34;noreferrer&#34;&gt;CVE-2026-31432&lt;/a&gt; and &lt;a href=&#34;https://nvd.nist.gov/vuln/detail/CVE-2026-31433&#34;  target=&#34;_blank&#34; rel=&#34;noreferrer&#34;&gt;CVE-2026-31433&lt;/a&gt;: two remote, unauthenticated OOB writes in the Linux kernel&amp;rsquo;s &lt;strong&gt;ksmbd&lt;/strong&gt;. Read on for the details of the setup that achieved this, including &amp;ndash; yes! &amp;ndash; getting LLMs drunk.&lt;/em&gt;&lt;/p&gt;&#xA;&#xA;&lt;h2 class=&#34;relative group&#34;&gt;Background&#xA;    &lt;div id=&#34;background&#34; class=&#34;anchor&#34;&gt;&lt;/div&gt;&#xA;    &#xA;    &lt;span&#xA;        class=&#34;absolute top-0 w-6 transition-opacity opacity-0 -start-6 not-prose group-hover:opacity-100 select-none&#34;&gt;&#xA;        &lt;a class=&#34;text-primary-300 dark:text-neutral-700 !no-underline&#34; href=&#34;#background&#34; aria-label=&#34;Anchor&#34;&gt;#&lt;/a&gt;&#xA;    &lt;/span&gt;&#xA;    &#xA;&lt;/h2&gt;&#xA;&lt;p&gt;&amp;ldquo;LLMing&amp;rdquo; vulnerability research has been on my &amp;ldquo;Do Something About This&amp;rdquo; list since DARPA&amp;rsquo;s &lt;a href=&#34;https://www.darpa.mil/research/programs/ai-cyber&#34;  target=&#34;_blank&#34; rel=&#34;noreferrer&#34;&gt;AIxCC&lt;/a&gt; and XBOW&amp;rsquo;s &lt;a href=&#34;https://xbow.com/blog/xbow-scoold-vuln&#34;  target=&#34;_blank&#34; rel=&#34;noreferrer&#34;&gt;initial results&lt;/a&gt;. But back in 2023-24, models required a lot of harnessing to get anything useful, tool use was rudimentary, and the idea of squeezing as much code as I could into a model&amp;rsquo;s context &amp;ndash; then triaging away the false positives &amp;ndash; filled me with dread.&lt;/p&gt;</description>
      <media:content xmlns:media="http://search.yahoo.com/mrss/" url="https://heyitsas.im/posts/drinking-llms/featured.webp" />
    </item>
    
    <item>
      <title>Spooler Alert: Remote Unauth&#39;d RCE-to-root Chain in CUPS</title>
      <link>https://heyitsas.im/posts/cups/</link>
      <pubDate>Sun, 05 Apr 2026 00:00:00 +0000</pubDate>
      
      <guid>https://heyitsas.im/posts/cups/</guid>
      <description>&lt;p&gt;&lt;em&gt;TLDR: my self-orchestrating team of &lt;a href=&#34;https://www.linkedin.com/posts/yasamal4ik_february-2026-cve-2026-26080-and-cve-2026-activity-7441018899502043136-Rukx&#34;  target=&#34;_blank&#34; rel=&#34;noreferrer&#34;&gt;vulnerability hunting agents&lt;/a&gt; discovered two issues in CUPS, &lt;a href=&#34;https://github.com/OpenPrinting/cups/security/advisories/GHSA-4852-v58g-6cwf&#34;  target=&#34;_blank&#34; rel=&#34;noreferrer&#34;&gt;CVE-2026-34980&lt;/a&gt; and &lt;a href=&#34;https://github.com/OpenPrinting/cups/security/advisories/GHSA-c54j-2vqw-wpwp&#34;  target=&#34;_blank&#34; rel=&#34;noreferrer&#34;&gt;CVE-2026-34990&lt;/a&gt;, chainable into &lt;code&gt;unauthenticated remote attacker -&amp;gt; unprivileged RCE -&amp;gt; root file (over)write&lt;/code&gt;. See below for the prerequisites, details, and mitigation options.&lt;/em&gt;&lt;/p&gt;&#xA;&#xA;&lt;h2 class=&#34;relative group&#34;&gt;Intro&#xA;    &lt;div id=&#34;intro&#34; class=&#34;anchor&#34;&gt;&lt;/div&gt;&#xA;    &#xA;    &lt;span&#xA;        class=&#34;absolute top-0 w-6 transition-opacity opacity-0 -start-6 not-prose group-hover:opacity-100 select-none&#34;&gt;&#xA;        &lt;a class=&#34;text-primary-300 dark:text-neutral-700 !no-underline&#34; href=&#34;#intro&#34; aria-label=&#34;Anchor&#34;&gt;#&lt;/a&gt;&#xA;    &lt;/span&gt;&#xA;    &#xA;&lt;/h2&gt;&#xA;&lt;p&gt;CUPS is &lt;em&gt;the&lt;/em&gt; standard way to do printing on Linux and other Unix(-like) systems. It&amp;rsquo;s been on my mind as a research target ever since doing incident response to Simone Margaritelli&amp;rsquo;s 2024 &lt;a href=&#34;https://www.evilsocket.net/2024/09/26/Attacking-UNIX-systems-via-CUPS-Part-I/&#34;  target=&#34;_blank&#34; rel=&#34;noreferrer&#34;&gt;unauth&amp;rsquo;d RCE finding&lt;/a&gt;, where he chained several CUPS vulnerabilities into an unauth&amp;rsquo;d RCE as &lt;code&gt;lp&lt;/code&gt;, the default CUPS service user.&lt;/p&gt;</description>
      <media:content xmlns:media="http://search.yahoo.com/mrss/" url="https://heyitsas.im/posts/cups/featured.webp" />
    </item>
    
  </channel>
</rss>
