Bug 16398 - Add Condition-Based CVE Reassessment Triggers to OE-Core
Summary: Add Condition-Based CVE Reassessment Triggers to OE-Core
Status: NEW
Alias: None
Product: OE-Core
Classification: Build System, Metadata & Runtime
Component: core (show other bugs)
Version: unspecified
Hardware: All Multiple
: Medium enhancement
Target Milestone: 6.1
Assignee: Eric Laurin
QA Contact:
URL:
Whiteboard:
Depends on:
Blocks:
 
Reported: 2026-08-17 19:15 UTC by Eric Laurin
Modified: 2026-08-20 18:26 UTC (History)
4 users (show)

See Also:
OS type for building Yocto: ---
Type of Regression: ---
Verified:
Documentation change: Don't know


Attachments
Markdown Files that describe the problem and proposed solution (9.90 KB, application/x-zip-compressed)
2026-08-17 19:15 UTC, Eric Laurin
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description Eric Laurin 2026-08-17 19:15:15 UTC
Created attachment 5242 [details]
Markdown Files that describe the problem and proposed solution

CVE dispositions can become invalid when a build configuration, kernel option, package version, image composition, or deployment condition changes, but OE-Core currently has no first-class way to record and evaluate those reassessment conditions.

This proposal introduces a declarative reassessment mechanism associated with CVE status entries. Maintainers could specify triggers such as configuration changes, kernel options becoming enabled, version thresholds, image/component changes, or elapsed review intervals. The CVE would then be automatically surfaced for reassessment when a relevant condition occurs.

This moves CVE maintenance beyond static suppressions and time-based review queues toward evidence-driven reassessment, reducing stale dispositions while preserving explicit human approval for any resulting metadata changes. The accompanying documentation includes the proposed syntax, trigger types, evaluation model, validation rules, compatibility considerations, and worked examples.

This is an early concept, and we do not have a working model or code. We are submitting this to determine if there is interest in something like this because we would require coaching and collaboration for such an effort but have the desire to participate in defining and implementing something that would help surface previously dispositioned CVEs through CVE_STATUS to identify when surrounding changes require the CVE_STATUS to be looked at.
Comment 1 Paul Barker 2026-08-20 14:39:05 UTC
Hi Eric, the intention here is good, but this is a classic case of over-design-by-AI upfront. There are ways to improve CVE_STATUS but you need to start with discussion, especially if you are unfamiliar with contributing in this area.
Comment 2 Eric Laurin 2026-08-20 18:26:36 UTC
Agreed, we are hoping for this to start a discussion. The intent may not even be correct because it propagates the situation of letting unpatched code linger. We have thought about why this happens even when there what should be sufficient resources in place to aggressively backport patches even when paying for support.

You are correct, I am not familiar with contributing, but willing to learn.