CVE-2026-53063: Linux dm-cache Write Hang in Passthrough Mode – Patch & Detection Guide
A flaw in the Linux kernel's device mapper cache subsystem causes write operations to hang when the cache is running in passthrough mode and invalidation occurs. The bug stems from incomplete logic in the invalidate_remove() function that sets up write requests but fails to submit them, leaving applications waiting indefinitely for I/O completion. This is a local issue requiring kernel privileges to trigger, but can severely disrupt system stability.
Source data · NVD / CISA · public domain
- CVSS
- 3.1 · 5.5 MEDIUM · CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H
- Weaknesses (CWE)
- —
- Affected products
- 1 configuration(s)
- Published / Modified
- 2026-06-24 / 2026-07-21
NVD description (verbatim)
In the Linux kernel, the following vulnerability has been resolved: dm cache: fix write hang in passthrough mode The invalidate_remove() function has incomplete logic for handling write hit bios after cache invalidation. It sets up the remapping for the overwrite_bio but then drops it immediately without submission, causing write operations to hang. Fix by adding a new invalidate_committed() continuation that submits the remapped writes to the cache origin after metadata commit completes, while using the overwrite_endio hook to ensure proper completion sequencing. This maintains existing coherency. Also improve error handling in invalidate_complete() to preserve the original error status instead of using bio_io_error() unconditionally.
6 reference(s) · View on NVD →
SEC.co analysis · AI-assisted, reviewed against source
Technical summary
CVE-2026-53063 affects the dm-cache (device mapper cache) module in the Linux kernel. The invalidate_remove() function handles write hits that occur after cache invalidation, but its implementation is incomplete: it remaps write bios for submission to the cache origin device but then drops them without actually submitting them to the I/O stack. This causes write operations to stall. The fix introduces a new invalidate_committed() continuation function that properly submits remapped writes after metadata commits complete, leveraging the overwrite_endio hook to maintain correct completion sequencing and preserve cache coherency. Additionally, error handling in invalidate_complete() is corrected to propagate the original error status instead of unconditionally invoking bio_io_error().
Business impact
Systems relying on dm-cache in passthrough mode for caching workloads will experience write hangs during invalidation operations, leading to application stalls and potential data pipeline disruptions. While this requires local access and elevated privileges to exploit, it can affect infrastructure using caching layers for performance optimization, particularly in virtualized or containerized deployments where dm-cache is used for tiered storage. Availability impact is significant; integrity and confidentiality are not directly affected.
Affected systems
This vulnerability affects the Linux kernel across all versions that include the dm-cache subsystem. Specifically, systems running Linux kernels with device mapper cache functionality configured, particularly in passthrough mode, are vulnerable. The issue manifests when cache invalidation is triggered during active write workloads. Both physical and virtual systems using dm-cache for storage optimization are in scope; the requirement for local access and elevated privileges (CAP_SYS_ADMIN or root) narrows the practical attack surface to local threat actors or privileged processes.
Exploitability
Exploitability is limited due to the requirement for local access and elevated privileges. This is not a remote vulnerability and cannot be triggered by unprivileged users. However, for privileged local processes, system administrators, or containers with excessive capabilities, triggering cache invalidation during write-heavy workloads is straightforward. The hang is deterministic once invalidation conditions are met, making it reliable for denial-of-service scenarios in multi-tenant environments or systems with weak privilege separation. No special tools or sophisticated techniques are required; standard cache management operations can trigger the condition.
Remediation
Apply the kernel patch that implements the invalidate_committed() continuation function and corrects error handling in invalidate_complete(). Verify the patch version against the official kernel security advisory or your distribution's update channel. For immediate mitigation in high-risk environments, disable passthrough mode on dm-cache instances if operationally feasible, or limit exposure by restricting access to cache management operations. Kernel upgrades from your Linux distribution should be prioritized; vendors typically backport fixes to supported stable branches.
Patch guidance
Obtain the patched kernel version from your Linux distribution's security updates or directly from kernel.org. Ubuntu, Red Hat, Debian, and other major distributions will release kernel updates addressing this flaw across supported releases. Verify that the update includes the fix to invalidate_remove() logic and the new invalidate_committed() continuation. Testing in a non-production environment before deployment is strongly recommended, particularly if dm-cache is actively caching critical workloads. Monitor for any regression in cache performance post-patch.
Detection guidance
Monitor system logs for evidence of stalled write operations on dm-cache backed devices, particularly if using passthrough mode. Watch for I/O timeouts, hung task warnings, or application-level write timeouts correlated with cache invalidation events. Kernel logs may show blocked or unkillable processes (D state) waiting on device I/O. Check dmesg output for dm-cache related messages around the time of suspected hangs. Network-based detection is not applicable; this requires direct access to affected systems. Correlate cache invalidation commands (dmsetup) with subsequent application performance degradation.
Why prioritize this
While the CVSS score of 5.5 (Medium) reflects the local-only nature of the vulnerability, the practical impact on affected systems can be severe. Any organization using dm-cache in production, particularly in tiered storage or caching acceleration scenarios, should prioritize patching to avoid write stalls that disrupt service availability. The fix is straightforward and unlikely to introduce regression, making this a high-confidence remediation candidate.
Risk score, explained
The CVSS v3.1 score of 5.5 (Medium) reflects an Attack Vector of Local, low Attack Complexity, requirement for Low privilege, no user interaction, impact limited to Availability (High), with no Confidentiality or Integrity impact. The score appropriately downgrades the severity below Critical or High due to privilege and access requirements, but the High Availability impact reflects the serious nature of write hangs in production systems.
Frequently asked questions
How do I know if my system uses dm-cache in passthrough mode?
Run 'dmsetup status' as root and look for cache targets configured with 'passthrough 1'. You can also check 'dmsetup table' to see cache device mapper configuration. If you are uncertain, consult your infrastructure or storage team about active caching policies.
Does this vulnerability affect caching modes other than passthrough?
The vulnerability specifically targets logic in passthrough mode handling. Write-back and write-through modes may have different code paths, but the fix addresses the incomplete invalidate_remove() function that operates during cache invalidation regardless of mode. Defer to the vendor advisory for precise affected configurations.
Can this be exploited remotely?
No. This requires local access and elevated privileges (CAP_SYS_ADMIN or root) to trigger cache invalidation. Remote users or unprivileged local users cannot exploit this vulnerability directly.
What should I do if I cannot patch immediately?
If feasible, disable passthrough mode on dm-cache instances or limit cache invalidation operations. Increase monitoring for write stalls and I/O timeouts. Implement access controls to restrict who can perform cache management operations. Plan and schedule patching as soon as feasible, as this is a local availability issue with a well-defined fix.
This analysis is based on the CVE description and CVSS vector provided. Specific patch versions, affected kernel releases, and distribution-specific guidance should be verified against official Linux vendor security advisories (Red Hat, Canonical, Debian, etc.) and kernel.org. Testing in non-production environments is mandatory before applying patches to production systems. Organizations should validate that published patches address the exact invalidate_committed() and error-handling changes described. SEC.co makes no warranty regarding the completeness or applicability of this guidance to specific deployments. Source: NVD (public-domain), retrieved 2026-08-01. Analysis generated by SEC.co (claude-haiku-4-5).
Affected vendors
Related vulnerabilities
- CVE-2025-71313MEDIUMLinux Kernel PCI Endpoint NULL Pointer Dereference
- CVE-2025-71314MEDIUMLinux Panthor GPU Driver Denial of Service via Cache Flush Timeout
- CVE-2025-71315MEDIUMLinux Kernel vkms DRM Vblank Timer Denial of Service
- CVE-2026-0268MEDIUMPrisma Access Agent Linux VPN Bypass Vulnerability
- CVE-2026-10004MEDIUMChrome UI Spoofing Vulnerability – Password Dialog Hijacking
- CVE-2026-10018MEDIUMInteger Overflow in Chrome ANGLE GPU Graphics Layer
- CVE-2026-10912MEDIUMChrome Extension Same-Origin Policy Bypass (CVSS 6.5)
- CVE-2026-10916MEDIUMChrome DevTools UXSS Vulnerability