CVE-2026-9640: LXD Privilege Escalation via Snapshot Policy Bypass in Multi-Tenant Environments
A vulnerability in LXD allows authenticated project operators to escalate their privileges to root on the host system by bypassing security policies designed to restrict multi-tenant environments. An attacker with operator-level access in a restricted project can craft a malicious instance backup, import it, and restore its snapshot in a way that applies restricted configuration settings that would normally be blocked. Once the instance restarts with these unauthorized settings, the operator gains full root access to the underlying host. This affects LXD versions 6.0 through 6.8, 5.21.0 through 5.21.4, and 5.0.0 through 5.0.6.
Source data · NVD / CISA · public domain
- CVSS
- 3.1 · 7.2 HIGH · CVSS:3.1/AV:N/AC:L/PR:H/UI:N/S:U/C:H/I:H/A:H
- Weaknesses (CWE)
- CWE-863
- Affected products
- 1 configuration(s)
- Published / Modified
- 2026-06-26 / 2026-07-02
NVD description (verbatim)
A privilege escalation vulnerability exists in LXD from 6.0 before 6.9, 5.21.0 before 5.21.5, and 5.0.0 before 5.0.7 regarding the handling of project-restriction policies during snapshot restoration.. An authenticated project operator in a restricted multi-tenant environment can bypass policy restrictions by importing a maliciously crafted instance backup containing restricted configuration keys within a snapshot. When the snapshot is restored, these restricted keys are applied to the live instance without policy validation. Starting the modified instance grants the operator unauthorized host root access.
5 reference(s) · View on NVD →
SEC.co analysis · AI-assisted, reviewed against source
Technical summary
The vulnerability stems from insufficient policy validation during snapshot restoration in LXD's project-restriction framework. When a project operator restores a snapshot from a backup, the system fails to re-validate restricted configuration keys that are embedded within the snapshot metadata. In a properly restricted multi-tenant environment, certain host-level configuration keys should be immutable and blocked from modification by project-scoped operators. However, this validation is bypassed during the snapshot restore operation, allowing an operator to inject prohibited keys into the instance configuration. Upon instance startup, these keys take effect without further policy checks, resulting in unauthorized access escalation to the host's root account. The root cause is inadequate input sanitization and policy enforcement in the snapshot restoration code path, classified as an authorization flaw (CWE-863).
Business impact
In multi-tenant LXD deployments used for containerized workloads, hosting, or cloud platforms, this vulnerability enables a compromised or malicious operator account to escape tenant isolation and compromise the entire host system. This breaks the fundamental trust boundary in shared infrastructure environments. An attacker gaining host root access can pivot to other tenant instances, exfiltrate sensitive data, modify billing systems, launch supply-chain attacks, or cause persistent service disruption. Organizations relying on LXD's project restrictions for security isolation face complete loss of that isolation layer if any operator account is compromised or acts maliciously.
Affected systems
The vulnerability affects Canonical LXD in three supported release series: versions 6.0 through 6.8 (inclusive), versions 5.21.0 through 5.21.4 (inclusive), and versions 5.0.0 through 5.0.6 (inclusive). LXD 6.9, 5.21.5, and 5.0.7 or later contain the fix. Organizations should verify their installed LXD version and cross-reference against these ranges. The vulnerability requires an authenticated account with project operator privileges, so it does not affect standalone or single-tenant LXD deployments without multi-project configurations, though those deployments may still be at risk if project features are enabled for operational flexibility.
Exploitability
Exploitation requires an existing authenticated project operator account in a restricted multi-tenant environment—an insider threat or compromised credential scenario. No network authentication is needed if the attacker already has operator access, and no user interaction is required to restore a snapshot. The attack surface is limited to authenticated users, but the barrier to entry is low once that access exists. The CVSS vector (AV:N/AC:L/PR:H) reflects network accessibility and high-privilege requirement; the high impact score (C:H/I:H/A:H) reflects complete host compromise. While this is not an unauthenticated remote code execution vulnerability, it is a severe privilege escalation that completely undermines the security model of multi-tenant LXD deployments.
Remediation
Upgrade LXD to version 6.9 or later for the LXD 6.x series, version 5.21.5 or later for the 5.21.x series, or version 5.0.7 or later for the 5.0.x series. These versions include policy validation fixes that prevent restricted keys from being restored without proper authorization checks. Additionally, conduct an audit of project operator accounts and restrict operator privileges to only those who require them. Consider reviewing snapshot import and restore workflows to ensure they originate from trusted, internal sources only. Organizations with shared-tenant infrastructure should prioritize this patch immediately.
Patch guidance
Apply the latest patch version from Canonical's LXD security advisories for your release series. Patch testing should be performed in a staging environment that mimics your multi-tenant configuration before production deployment. Because this vulnerability affects the core tenant-isolation mechanism, patching should be treated as critical security maintenance. Plan patching as soon as feasible rather than deferring to a standard maintenance window. Canonical provides detailed upgrade instructions in their release notes; verify compatibility with your deployment topology and any custom scripts or integrations before applying.
Detection guidance
Monitor for suspicious snapshot import and restore operations, particularly when performed by project operators on instances that do not normally use snapshots. Log analysis should flag cases where a snapshot restoration precedes unusual instance configuration changes or host-level access patterns. Intrusion detection can watch for unexpected privilege escalation sequences following snapshot operations. Additionally, audit the project operator role memberships and verify they align with organizational access policies. If LXD audit logging is enabled, review logs for snapshot operations combined with subsequent host access attempts. Organizations can also validate the running configuration of critical instances against baseline snapshots to detect unauthorized changes.
Why prioritize this
This vulnerability merits immediate prioritization in multi-tenant LXD environments because it directly breaks the tenant-isolation model that these deployments rely on for security. A single compromised operator account can escalate to full host compromise, affecting all co-hosted tenants and the underlying infrastructure. The CVSS score of 7.2 (HIGH) reflects the severity of unauthorized root access, though the true business impact in shared environments is even more critical. The patch is available and relatively straightforward to deploy, making remediation both urgent and feasible.
Risk score, explained
The CVSS 3.1 score of 7.2 reflects a high-severity vulnerability with network accessibility, low attack complexity, and high-privilege requirement, coupled with complete impact across confidentiality, integrity, and availability. In the context of multi-tenant LXD deployments, the actual risk is amplified because a single operator escalation undermines the entire tenant-isolation architecture, affecting multiple customers or business units simultaneously. Organizations not using LXD's multi-project/multi-tenant features face lower risk, but those relying on project restrictions for security should treat this as a critical vulnerability requiring emergency patching. The vulnerability is not currently listed on CISA's Known Exploited Vulnerabilities catalog, but the low barrier to exploitation (requiring only authenticated access) and high impact (host root access) mean it could be weaponized quickly if disclosed publicly.
Frequently asked questions
Does this vulnerability affect single-tenant or standalone LXD deployments?
The vulnerability requires a project operator in a restricted multi-tenant environment. Single-tenant deployments or those not using LXD's project-restriction features are not affected by this specific privilege escalation. However, if you have any multi-project configuration enabled for operational flexibility, you should still patch to eliminate the risk.
What is the difference between a project operator and other LXD user roles?
A project operator has administrative privileges within a specific LXD project but is restricted by project policies from modifying host-level settings. This vulnerability allows such an operator to bypass those policy restrictions and achieve unauthorized host root access. Regular project users without operator status cannot exploit this vulnerability.
Can we detect if this vulnerability has been exploited in our environment?
Detection requires examining LXD logs and system audit trails for snapshot import/restore operations combined with unusual instance configuration changes or host-level access patterns. Compare current instance configurations against known-good baselines to identify unauthorized modifications. However, detection may be difficult if audit logging was not previously enabled, so proactive patching is strongly recommended.
What should we do if we discover unauthorized snapshot imports or restores?
Immediately isolate affected instances, review host access logs for the time period following the suspicious snapshot operation, and check all co-hosted tenants for signs of compromise. Perform a forensic analysis to determine what configuration changes were applied and what access was gained. Once the host and all tenants are confirmed clean, patch LXD and review operator role assignments to prevent recurrence.
This analysis is based on published vulnerability information and CVE data current as of the stated publication date. Specific patch version numbers, compatibility details, and Canonical's security advisories should be verified directly with LXD's official release notes and security channels before implementing remediation. This vulnerability requires pre-existing authenticated project operator access; it does not enable unauthenticated remote code execution. Organizations should assess their specific LXD deployment topology and multi-tenancy configuration to determine actual risk level. No proof-of-concept code is provided; exploitation should not be attempted outside controlled, authorized testing environments. Source: NVD (public-domain), retrieved 2026-08-05. Analysis generated by SEC.co (claude-haiku-4-5).
Related vulnerabilities
- CVE-2016-20075HIGHWordPress Ultimate Product Catalog 3.8.6 Arbitrary File Upload (CVSS 8.8)
- CVE-2025-14774HIGHABB T-MAC Plus Denial-of-Service Vulnerability (CVSS 7.4)
- CVE-2025-32348HIGHAndroid Local Privilege Escalation via Missing Permission Check
- CVE-2026-0272HIGHPalo Alto PAN-OS Privilege Escalation Vulnerability (PA-Series, VM-Series, Panorama)
- CVE-2026-14536HIGHDevolutions Server MFA Bypass – High Severity Authentication Flaw
- CVE-2026-15125HIGHChrome Forms Code Execution Vulnerability
- CVE-2026-21031HIGHAppBlock Authorization Flaw in Samsung Android—Risk & Patch Guidance
- CVE-2026-22659HIGHFlaskBB Authorization Bypass in Batch Topic Operations