<?xml version='1.0' encoding='UTF-8'?>
<?xml-stylesheet href="/static/style.xsl" type="text/xsl"?>
<rss xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/" version="2.0">
  <channel>
    <title>Most recent entries from all</title>
    <link>https://cve.radiocsirt.org</link>
    <description>Contains only the most 10 recent entries.</description>
    <docs>http://www.rssboard.org/rss-specification</docs>
    <generator>python-feedgen</generator>
    <language>en</language>
    <lastBuildDate>Tue, 17 Nov 2026 08:04:27 +0000</lastBuildDate>
    <item>
      <title>EUVD-2026-392531</title>
      <link>https://cve.radiocsirt.org/vuln/euvd-2026-392531</link>
      <description>EUVD-2026-392531</description>
      <guid isPermaLink="false">https://cve.radiocsirt.org/vuln/euvd-2026-392531</guid>
    </item>
    <item>
      <title>fkie_cve-2026-115921</title>
      <link>https://cve.radiocsirt.org/vuln/fkie_cve-2026-115921</link>
      <description>&lt;p&gt;Open Neural Network Exchange (ONNX) is an open standard for machine learning interoperability. From 1.4.0 until 1.22.0, ScanInferenceFunction, ScanInferenceFunction_opset8, and ScanInferenceFunction_opset9 do not safely validate the required num_scan_inputs attribute. A missing attribute is dereferenced as a null pointer, while a negative value is converted to size_t and can reach an exceptional oversized allocation path. An untrusted model can therefore terminate or disrupt shape inference, including a negative-value model that can pass onnx.checker.check_model(). This issue is fixed in version 1.22.0.&lt;/p&gt;</description>
      <content:encoded>&lt;p&gt;Open Neural Network Exchange (ONNX) is an open standard for machine learning interoperability. From 1.4.0 until 1.22.0, ScanInferenceFunction, ScanInferenceFunction_opset8, and ScanInferenceFunction_opset9 do not safely validate the required num_scan_inputs attribute. A missing attribute is dereferenced as a null pointer, while a negative value is converted to size_t and can reach an exceptional oversized allocation path. An untrusted model can therefore terminate or disrupt shape inference, including a negative-value model that can pass onnx.checker.check_model(). This issue is fixed in version 1.22.0.&lt;/p&gt;</content:encoded>
      <guid isPermaLink="false">https://cve.radiocsirt.org/vuln/fkie_cve-2026-115921</guid>
    </item>
    <item>
      <title>GHSA-m4pr-6v3c-6fgr — ONNX: Null Pointer Dereference in ScanInferenceFunction Due to Missing Required Attribute</title>
      <link>https://cve.radiocsirt.org/vuln/ghsa-m4pr-6v3c-6fgr</link>
      <description>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: onnx&lt;/p&gt;
&lt;p&gt;## Summary&lt;/p&gt;
&lt;p&gt;ONNX shape inference did not safely validate the required Scan `num_scan_inputs` attribute. Two malformed values could terminate or disrupt the process:&lt;/p&gt;
&lt;p&gt;1. If the attribute was missing, `ctx.getAttribute(&amp;#34;num_scan_inputs&amp;#34;)` returned null and was dereferenced.
2. If the attribute was negative, conversion to `size_t` produced a very large value that reached a vector allocation and raised `std::length_error`.&lt;/p&gt;
&lt;p&gt;The negative-value variant passed `onnx.checker.check_model()` before reaching shape inference.&lt;/p&gt;
&lt;p&gt;## Technical details&lt;/p&gt;
&lt;p&gt;The vulnerable Scan inference implementations used the following pattern:&lt;/p&gt;
&lt;p&gt;```cpp
auto num_scan_inputs =
    narrow_cast&amp;lt;size_t&amp;gt;(ctx.getAttribute(&amp;#34;num_scan_inputs&amp;#34;)-&amp;gt;i());
```&lt;/p&gt;
&lt;p&gt;This code neither checked the returned pointer nor rejected a negative signed value before converting it to an unsigned size. The pattern was present in the current Scan inference function and the older opset 8 and 9 paths.&lt;/p&gt;
&lt;p&gt;The fix uses the required-attribute helper and checked narrowing:&lt;/p&gt;
&lt;p&gt;```cpp
auto num_scan_inputs =
    narrow&amp;lt;size_t&amp;gt;(getRequiredAttributeInt(ctx, &amp;#34;num_scan_inputs&amp;#34;));
```&lt;/p&gt;
&lt;p&gt;## Impact&lt;/p&gt;
&lt;p&gt;An application that performs shape inference on an untrusted model can be terminated or forced into an exceptional allocation path, causing denial of service. No confidentiality or integrity impact was demonstrated.&lt;/p&gt;
&lt;p&gt;## Affected versions&lt;/p&gt;
&lt;p&gt;- Affected: ONNX 1.4.0 through 1.21.0
- Patched: ONNX 1.22.0&lt;/p&gt;
&lt;p&gt;ONNX 1.4.0 is the first release containing the vulnerable Scan shape-in…&lt;/p&gt;</description>
      <content:encoded>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: onnx&lt;/p&gt;
&lt;p&gt;## Summary&lt;/p&gt;
&lt;p&gt;ONNX shape inference did not safely validate the required Scan `num_scan_inputs` attribute. Two malformed values could terminate or disrupt the process:&lt;/p&gt;
&lt;p&gt;1. If the attribute was missing, `ctx.getAttribute(&amp;#34;num_scan_inputs&amp;#34;)` returned null and was dereferenced.
2. If the attribute was negative, conversion to `size_t` produced a very large value that reached a vector allocation and raised `std::length_error`.&lt;/p&gt;
&lt;p&gt;The negative-value variant passed `onnx.checker.check_model()` before reaching shape inference.&lt;/p&gt;
&lt;p&gt;## Technical details&lt;/p&gt;
&lt;p&gt;The vulnerable Scan inference implementations used the following pattern:&lt;/p&gt;
&lt;p&gt;```cpp
auto num_scan_inputs =
    narrow_cast&amp;lt;size_t&amp;gt;(ctx.getAttribute(&amp;#34;num_scan_inputs&amp;#34;)-&amp;gt;i());
```&lt;/p&gt;
&lt;p&gt;This code neither checked the returned pointer nor rejected a negative signed value before converting it to an unsigned size. The pattern was present in the current Scan inference function and the older opset 8 and 9 paths.&lt;/p&gt;
&lt;p&gt;The fix uses the required-attribute helper and checked narrowing:&lt;/p&gt;
&lt;p&gt;```cpp
auto num_scan_inputs =
    narrow&amp;lt;size_t&amp;gt;(getRequiredAttributeInt(ctx, &amp;#34;num_scan_inputs&amp;#34;));
```&lt;/p&gt;
&lt;p&gt;## Impact&lt;/p&gt;
&lt;p&gt;An application that performs shape inference on an untrusted model can be terminated or forced into an exceptional allocation path, causing denial of service. No confidentiality or integrity impact was demonstrated.&lt;/p&gt;
&lt;p&gt;## Affected versions&lt;/p&gt;
&lt;p&gt;- Affected: ONNX 1.4.0 through 1.21.0
- Patched: ONNX 1.22.0&lt;/p&gt;
&lt;p&gt;ONNX 1.4.0 is the first release containing the vulnerable Scan shape-in…&lt;/p&gt;</content:encoded>
      <guid isPermaLink="false">https://cve.radiocsirt.org/vuln/ghsa-m4pr-6v3c-6fgr</guid>
    </item>
    <item>
      <title>msrc_CVE-2026-115921 — ONNX: Null Pointer Dereference in ScanInferenceFunction Due to Missing Required Attribute</title>
      <link>https://cve.radiocsirt.org/vuln/msrc_cve-2026-115921</link>
      <description>msrc_CVE-2026-115921</description>
      <guid isPermaLink="false">https://cve.radiocsirt.org/vuln/msrc_cve-2026-115921</guid>
    </item>
    <item>
      <title>UBUNTU-CVE-2026-115921</title>
      <link>https://cve.radiocsirt.org/vuln/ubuntu-cve-2026-115921</link>
      <description>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; Ubuntu:22.04:LTS: onnx, Ubuntu:Pro:24.04:LTS: onnx, Ubuntu:26.10: onnx, Ubuntu:26.04:LTS: onnx&lt;/p&gt;
&lt;p&gt;Open Neural Network Exchange (ONNX) is an open standard for machine learning interoperability. From 1.4.0 until 1.22.0, ScanInferenceFunction, ScanInferenceFunction_opset8, and ScanInferenceFunction_opset9 do not safely validate the required num_scan_inputs attribute. A missing attribute is dereferenced as a null pointer, while a negative value is converted to size_t and can reach an exceptional oversized allocation path. An untrusted model can therefore terminate or disrupt shape inference, including a negative-value model that can pass onnx.checker.check_model(). This issue is fixed in version 1.22.0.&lt;/p&gt;</description>
      <content:encoded>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; Ubuntu:22.04:LTS: onnx, Ubuntu:Pro:24.04:LTS: onnx, Ubuntu:26.10: onnx, Ubuntu:26.04:LTS: onnx&lt;/p&gt;
&lt;p&gt;Open Neural Network Exchange (ONNX) is an open standard for machine learning interoperability. From 1.4.0 until 1.22.0, ScanInferenceFunction, ScanInferenceFunction_opset8, and ScanInferenceFunction_opset9 do not safely validate the required num_scan_inputs attribute. A missing attribute is dereferenced as a null pointer, while a negative value is converted to size_t and can reach an exceptional oversized allocation path. An untrusted model can therefore terminate or disrupt shape inference, including a negative-value model that can pass onnx.checker.check_model(). This issue is fixed in version 1.22.0.&lt;/p&gt;</content:encoded>
      <guid isPermaLink="false">https://cve.radiocsirt.org/vuln/ubuntu-cve-2026-115921</guid>
    </item>
  </channel>
</rss>
