PatchSiren cyber security CVE debrief
CVE-2026-63950 Linux CVE debrief
A vulnerability was found in the Linux kernel's memory management subsystem, specifically in the try_to_unmap_one function. The bug occurs due to the improper initialization of the nr_pages variable, which can lead to potential refcount/mapcount corruption when a 64K large folio is mmaped with MAP_ANONYMOUS | MAP_DROPPABLE, and then madvise(MADV_FREE) is called, followed by making the last page device-exclusive via HMM_DMIRROR_EXCLUSIVE. This issue was resolved by initializing nr_pages to 1 at the start of each loop iteration in try_to_unmap_one. The vulnerability was introduced by lazyfree folio batching, and device-exclusive only works for anonymous folios. Linux kernel developers, maintainers, and users who rely on the Linux kernel for their systems should be aware of this issue and take necessary actions.
- Vendor
- Linux
- Product
- Unknown
- CVSS
- Unknown
- CISA KEV
- Not listed in stored evidence
- Original CVE published
- 2026-07-19
- Original CVE updated
- 2026-07-19
- Advisory published
- 2026-07-19
- Advisory updated
- 2026-07-19
Who should care
Linux kernel developers, maintainers, and users who rely on the Linux kernel for their systems should be aware of this issue and take necessary actions to apply the official patch and review the patched Linux kernel version.
Technical summary
The vulnerability is caused by the improper initialization of the nr_pages variable in the try_to_unmap_one function. This can lead to refcount/mapcount corruption when a 64K large folio is mmaped with MAP_ANONYMOUS | MAP_DROPPABLE, and then madvise(MADV_FREE) is called, followed by making the last page device-exclusive via HMM_DMIRROR_EXCLUSIVE. The bug was introduced by lazyfree folio batching, and device-exclusive only works for anonymous folios. The issue was resolved by initializing nr_pages to 1 at the start of each loop iteration in try_to_unmap_one.
Defensive priority
High
Recommended defensive actions
- Apply the official patch to initialize nr_pages to 1 at loop start in try_to_unmap_one
- Review and test the patched Linux kernel version
- Monitor system logs for potential issues related to folio_referenced_one and try_to_unmap_one
- Confirm whether affected product deployments exist in managed environments and assign an owner for follow-up.
- Review the supplied official advisory or CVE record to validate affected scope, severity, and vendor guidance.
- Plan vendor-supported updates or mitigations through normal change control where exposure is confirmed.
- Check relevant monitoring, detection, and logs for exposed assets that need extra review.
Evidence notes
The vulnerability was resolved by initializing nr_pages to 1 at the start of each loop iteration in try_to_unmap_one. The bug was introduced by lazyfree folio batching, and device-exclusive only works for anonymous folios. The userspace visible effect is simply kernel crashing somewhere due to refcount/mapcount corruption. The source confidence is limited, and defenders should verify the patched Linux kernel version.
Official resources
-
CVE-2026-63950 CVE record
CVE.org
-
CVE-2026-63950 NVD detail
NVD
-
Source item URL
nvd_modified
-
Source reference
416baaa9-dc9f-4396-8d5f-8c081fb06d67
-
Source reference
416baaa9-dc9f-4396-8d5f-8c081fb06d67
-
Source reference
416baaa9-dc9f-4396-8d5f-8c081fb06d67
AI-assisted PatchSiren debrief based on the supplied source corpus. The CVE record was published on 2026-07-19T16:17:13.730Z and has not been modified since then.