GHSA-rgg9-956j-3378

Suggest an improvement
Source
https://github.com/advisories/GHSA-rgg9-956j-3378
Import Source
https://github.com/github/advisory-database/blob/main/advisories/github-reviewed/2026/11/GHSA-rgg9-956j-3378/GHSA-rgg9-956j-3378.json
JSON Data
https://api.osv.dev/v1/vulns/GHSA-rgg9-956j-3378
Aliases
Published
2026-11-19T16:04:43Z
Modified
2026-11-19T16:15:03Z
Severity
  • 6.8 (Medium) CVSS_V4 - CVSS:4.0/AV:L/AC:L/AT:P/PR:L/UI:N/VC:N/VI:H/VA:N/SC:N/SI:H/SA:N CVSS Calculator
Summary
Mareforma: invalidated claims still promote to REPLICATED
Details

What's wrong

_maybe_update_replicated_unlocked never checks t_invalid, so a claim that a signed contradiction verdict has marked invalid can still climb to REPLICATED through convergence detection, and it pulls an honest peer up with it.

Three gaps line up:

  • The new-claim short-circuit reads only status and asserter_keyid (core.py:1339-1352); an invalidated but still-open new claim passes.
  • The candidate-peer SELECT filters status = 'open', unresolved = 0, and transparency_logged = 1 but has no t_invalid predicate (core.py:1390-1410), so an invalidated existing claim qualifies as a convergence partner.
  • The promotion UPDATE gates only on status = 'open' (core.py:1422-1426), leaving a TOCTOU window even after the SELECT is fixed.

A signed contradiction verdict is meant to be terminal for the trust ladder: validate_claim refuses to promote a t_invalid claim to ESTABLISHED, and record_replication_verdict already gates its promotion UPDATE on t_invalid IS NULL. Convergence promotion is the one path that ignores the verdict. Reproduced at runtime: contradict claim X (t_invalid set, still PRELIMINARY), then assert peer Y sharing X's ESTABLISHED anchor; both X and Y land at REPLICATED, with X's t_invalid still set.

Severity and impact

High. A refuted claim re-enters the trust ladder at REPLICATED after its contradiction, and it serves as the corroborating peer that promotes an honest claim. Both outcomes contradict the documented refutation model, and REPLICATED is the input to further validation.

Location

  • mareforma/db/core.py:1339 new-claim short-circuit checks status and asserter_keyid but not t_invalid
  • mareforma/db/core.py:1390 candidate-peer SELECT WHERE clause has no t_invalid IS NULL predicate
  • mareforma/db/core.py:1422 promotion UPDATE gates on status = 'open' only, reopening the window between SELECT and UPDATE

Fix direction

Treat t_invalid the way the neighboring code already does. Extend the new-claim short-circuit to also read t_invalid and return when it is set. Add AND c.t_invalid IS NULL to the candidate-peer SELECT. Add AND t_invalid IS NULL to the promotion UPDATE's WHERE clause so the row-level lock closes the TOCTOU window, matching the gate record_replication_verdict already applies at core.py:3539.

Do red/green TDD: add a regression test that fails on main, then make it pass.

Quality bar

SOLID, DRY, KISS. Clean, readable, well-tested code. No new errors, warnings, or test failures; the full suite stays green on all supported Python versions.

Database specific
{
    "cwe_ids": [
        "CWE-754"
    ],
    "github_reviewed": true,
    "github_reviewed_at": "2026-11-19T16:04:43Z",
    "nvd_published_at": "2026-11-18T23:17:21Z",
    "severity": "MODERATE"
}
References

Affected packages

PyPI / mareforma

Package

Affected ranges

Type
ECOSYSTEM
Events
Introduced
0.3.0
Fixed
0.3.8

Affected versions

0.*
0.3.0
0.3.1
0.3.2
0.3.3
0.3.4
0.3.5
0.3.6
0.3.7

Database specific

source
"https://github.com/github/advisory-database/blob/main/advisories/github-reviewed/2026/11/GHSA-rgg9-956j-3378/GHSA-rgg9-956j-3378.json"