CVE-2026-97945

UNRATEDCVSS · not yet scoredTrending — 6 sources updated this week
—
EchelonGraph verdictMonitorLow exploitation likelihood right now — keep watching.
  • No CVSS published and no exploitation signals yet
CISA-KEV: Not listedEPSS PROB: 0.2%CVSS v2: —Exploit: None knownExposed services: Not assessed

No fix is confirmed yet — apply a workaround or compensating control (WAF / firewall / segmentation) and watch for the fix.

In the Linux kernel, the following vulnerability has been resolved:

x86/mm: Fix user-space data loss with MADV_FREE and THP

Some of users of Polars (a data analytics library) have lost production data from this bug. They seem to have just the right combination of huge pages, MADV_FREE and heavy reclaim pressure.

pmd_modify() masks the old value with (_HPAGE_CHG_MASK & ~_PAGE_DIRTY), silently discarding the hardware dirty bit. The subsequent pmd_mksaveddirty() call is supposed to transfer _PAGE_DIRTY into _PAGE_SAVED_DIRTY when write-protecting, but the dirty bit was already stripped from the value, so there is nothing left to transfer.

Contrast with pte_modify(), which keeps _PAGE_DIRTY_BITS in its mask, and pud_modify(), which keeps _HPAGE_CHG_MASK untouched: pmd_modify() is the odd one out. Any pmd_modify() on a writable, dirty PMD loses the dirty state.

One visible consequence is data loss with MADV_FREE on PMD-mapped THP:

memset(buf, 0x5A, size); // PMD-mapped THP, PMD dirty madvise(buf, size, MADV_FREE); // PMD cleaned but left writable, // folio marked lazyfree memset(buf, 0x5A, size); // hardware sets _PAGE_DIRTY again mprotect(buf, size, PROT_READ); // pmd_modify() drops the dirty bit mprotect(buf, size, PROT_READ|PROT_WRITE); // ... memory pressure ...

Reclaim (e.g. under memcg pressure) then finds the lazyfree folio with no dirty bit set anywhere and frees it in __discard_anon_folio_pmd_locked(), even though the data was rewritten after MADV_FREE; subsequent reads fault in fresh zero pages. NUMA hinting alone can trigger the same loss, as do_huge_pmd_numa_page() restores the PMD through pmd_modify() as well.

PMD-mapped file THPs are affected too: mprotect()/NUMA hinting dropping the dirty bit means rewritten data is never written back.

Fix it by keeping _PAGE_DIRTY in the preserved mask, exactly like pte_modify() and pud_modify() do. The existing pmd_mksaveddirty()/pmd_clear_saveddirty() pair then performs the hardware-dirty <-> saved-dirty transition based on the write bit, preserving the shadow-stack encoding rules.

CVSS v3
—
EchelonGraph score
Not yet assessedNo source has published severity data for this CVE yet — no CVSS score from NVD or a CNA, no GitHub advisory, and it is not in CISA KEV. This is not a rating of zero; we cannot assess it yet.
EG Score
—
EG Risk
—
EPSS PROB
0.2%
EPSS %ILE
8th
KEV
Not listed

Published

September 25, 2026

Last Modified

October 3, 2026

Vendor Advisories for CVE-2026-97945(2)

These vendors published their own advisory mentioning this CVE — often with vendor-specific remediation steps + affected product lists not in NVD.

Affected Packages

(3 across 3 ecosystems)
Debian:12(1)
PackageVulnerable rangeFix by version rangeDependents
linux-6.126.12.100-1~deb12u1, 6.12.101-1~deb12u1, 6.12.107-1~deb12u1
  • every version up to 6.12.111-1~deb12u1: fixed in 6.12.111-1~deb12u1
—
Debian:13(1)
PackageVulnerable rangeFix by version rangeDependents
linux6.12.100-1 ... 6.12.96-1 (35 versions)
  • every version up to 6.12.111-1: fixed in 6.12.111-1
—
Debian:14(1)
PackageVulnerable rangeFix by version rangeDependents
linux6.12.100-1 ... 7.2~rc7-1~exp1 (180 versions)
  • every version up to 7.2.7-1: fixed in 7.2.7-1
—

Data Freshness Timeline

(refreshed 19× in last 7d / 24× in last 30d)

Each row is a source pipeline that fetched or updated this CVE on that date, with what changed. For example, "NVD update" means NVD published or revised its analysis for this CVE; "MITRE cvelistV5" means we ingested or refreshed it from the CNA feed. Most recent first.

  1. 2026-10-04 23:23 UTCEPSS rescore
  2. 2026-10-03 14:26 UTCEPSS rescore
  3. 2026-10-03 11:29 UTCVendor advisory
  4. 2026-10-03 11:29 UTCGHSA enrichment
  5. 2026-10-03 11:26 UTCNVD update
  6. 2026-10-03 11:17 UTCVendor advisory
  7. 2026-10-03 11:17 UTCGHSA enrichment
  8. 2026-10-03 11:10 UTCMITRE cvelistV5
  9. 2026-10-02 17:57 UTCEPSS rescore
  10. 2026-10-01 19:51 UTCEPSS rescore
  11. 2026-10-01 14:58 UTCEG score recompute
  12. 2026-10-01 14:58 UTCVendor advisory
  13. 2026-10-01 14:58 UTCGHSA enrichment
  14. 2026-09-30 15:04 UTCEPSS rescore
  15. 2026-09-28 13:52 UTCEPSS rescore
  16. 2026-09-28 13:52 UTCEPSS rescore
  17. 2026-09-28 13:17 UTCEG score recompute
  18. 2026-09-28 13:17 UTCVendor advisory
  19. 2026-09-28 13:17 UTCGHSA enrichment
  20. 2026-09-27 13:49 UTCEPSS rescore
  21. 2026-09-26 15:59 UTCEPSS rescore
  22. 2026-09-25 11:30 UTCNVD update
  23. 2026-09-25 10:33 UTCEG score recompute
  24. 2026-09-25 10:27 UTCMITRE cvelistV5first tracked

Frequently asked(4)

What is CVE-2026-97945?
CVE-2026-97945 is a publicly disclosed vulnerability published on September 25, 2026. In the Linux kernel, the following vulnerability has been resolved: x86/mm: Fix user-space data loss with MADV_FREE and THP Some of users of Polars (a data analytics library) have lost production data from this bug. They seem to have just the right combination of huge pages, MADV_FREE and heavy…
When was CVE-2026-97945 disclosed?
CVE-2026-97945 was first published on September 25, 2026, with the most recent update on October 3, 2026. EchelonGraph re-ingests CVE updates from NVD on a 2-hour cycle, so this page reflects the latest published state.
Is CVE-2026-97945 actively exploited?
CVE-2026-97945 is not currently on CISA's Known Exploited Vulnerabilities catalog. FIRST EPSS estimates a 0.2% probability of exploitation in the next 30 days (8th percentile of EPSS-scored CVEs).
How do I remediate CVE-2026-97945?
No fix for CVE-2026-97945 is confirmed yet. Until one is published, restrict network exposure of the affected system or apply the vendor's mitigation — for example, keep it off the internet or limit it to trusted networks — and watch the vendor's advisory for the fix. The vendor advisories EchelonGraph has for CVE-2026-97945 are linked in the Vendor Advisories panel on this page.

Dependency Blast Radius

See which npm, PyPI, Go, and Maven packages are affected by CVE-2026-97945

Explore →

Is Your Infrastructure Affected by CVE-2026-97945?

EchelonGraph automatically scans your cloud infrastructure and maps CVE exposure using blast radius analysis.