CVE-2026-61711
Docker 脆弱性の分析と軽減

概要

CVE-2026-61711 is a security feature bypass vulnerability in Moby BuildKit, a toolkit for converting source code to build artifacts. A custom frontend could place an invalid SecurityMode value in a crafted build request, causing executor/oci/spec_linux.go to treat the unsupported value as a non-sandbox mode without requiring the security.insecure entitlement — thereby disabling Seccomp and AppArmor protections for the build container. All versions up to and including v0.31.0 are affected; the issue was fixed in v0.31.1. It carries a CVSS v4.0 base score of 5.3 (Moderate/Medium) (Github Advisory, BuildKit Release).

技術的な詳細

The root cause is improper input validation (CWE-20) in the generateSecurityOpts function within executor/oci/spec_linux.go. Prior to the fix, the function used a switch statement that only explicitly handled SecurityMode_INSECURE and SecurityMode_SANDBOX; any other integer value fell through without applying sandbox security options (Seccomp and AppArmor), effectively treating it as an insecure mode without checking for the security.insecure entitlement. An attacker with the ability to submit a custom frontend build request over the network (with low privileges) could craft a build request setting SecurityMode to an invalid enum value (e.g., pb.SecurityMode(2)) to trigger this path. The fix introduced a ValidateSecurityMode() function in solver/pb/securitymode.go that rejects any value other than SecurityMode_SANDBOX or SecurityMode_INSECURE, applied consistently across all platform-specific spec files (Github Advisory, Fix Commit).

影響

Successful exploitation reduces the isolation of the affected build container by disabling Seccomp and AppArmor kernel security profiles, allowing the container to perform syscalls or operations that would otherwise be blocked by those profiles. Notably, Linux capabilities remain restricted, limiting the severity of the bypass. An attacker exploiting this vulnerability could potentially execute restricted syscalls within the build container, access sensitive information, or perform unauthorized operations that the Seccomp/AppArmor profiles were designed to prevent — though full container escape is not directly implied (Github Advisory, Feedly).

エクスプロイト可能性

There is no public proof-of-concept exploit and no evidence of in-the-wild exploitation at this time (Feedly). The vulnerability is not listed in the CISA Known Exploited Vulnerabilities (KEV) catalog. The EPSS score is approximately 0.357% (29th percentile), indicating a low probability of exploitation in the near term. Exploitation requires low-level privileges (ability to submit custom frontend build requests to a BuildKit instance) and is not automatable according to NVD SSVC assessment.

エクスプロイテーションのステップ

  1. Identify a target: Locate a BuildKit instance (v0.31.0 or earlier) accessible over the network that accepts custom frontend build requests.
  2. Craft a malicious build request: Construct a build request using the BuildKit client API or gateway interface, setting the SecurityMode field in an ExecOp or StartRequest to an invalid integer value not defined in the SecurityMode enum (e.g., pb.SecurityMode(2)).
  3. Submit the request: Send the crafted request to the BuildKit daemon. Because the invalid value is not validated, generateSecurityOpts in spec_linux.go falls through without applying Seccomp or AppArmor profiles.
  4. Execute restricted operations: Within the resulting build container — which now lacks Seccomp and AppArmor enforcement — execute syscalls or operations that would normally be blocked by those profiles, potentially accessing sensitive data or performing unauthorized actions (Github Advisory, Fix Commit).

妥協の兆候

  • Logs: BuildKit daemon logs showing build requests from custom or untrusted frontends with unusual or unexpected SecurityMode values; absence of Seccomp-related log entries for containers that should have sandbox protections applied.
  • Process: Build containers where /proc/self/status shows Seccomp: 0 (Seccomp disabled) despite not having the security.insecure entitlement explicitly granted.
  • Network: Unexpected or unauthorized connections to the BuildKit gRPC API endpoint from unknown or untrusted frontend images.

軽減策と回避策

Upgrade BuildKit to version v0.31.1 or later, which introduces ValidateSecurityMode() validation that rejects any unknown SecurityMode values before generating executor specs (BuildKit Release). As a workaround for environments that cannot immediately upgrade, restrict the ability to submit custom frontend build requests to trusted users and trusted frontend images only (Github Advisory). Additionally, consider implementing supplementary runtime security controls (e.g., host-level Seccomp policies or mandatory access control) to compensate for the weakened container isolation until patching is complete.

コミュニティの反応

The vulnerability was reported by security researcher Alex0Young and fixed by BuildKit maintainer Tõnis Tiigi. It was released as part of a security patch release (v0.31.1) that also addressed a separate low-severity runtime DoS issue (GHSA-72x6-4j93-7w86). The advisory was rated "Low" severity by the maintainers in the repository advisory, though the GitHub Advisory Database classifies it as "Moderate" based on the CVSS v4.0 score of 5.3 (Github Advisory, BuildKit Release).

関連情報

Linuxディストリビューションの修正状況

主要なLinuxディストリビューションおよびそのリリースにおける修正の提供状況。

RHEL / CentOS

影響

OpenShift

openshift4/cnf-tests-rhel8

影響

RHEL 8

影響なし

RHEL 9

ubi9/buildah

影響

RHEL 10

rhel10-eus/rhel-10.0-bootc

影響

ソースこのレポートは AI を使用して生成されました

関連 Docker 脆弱 性:

CVE 識別子

重大度

スコア

テクノロジー

コンポーネント名

CISA KEV エクスプロイト

修正あり

公開日

CVE-2026-78662HIGH7.5
  • Docker logoDocker
  • headlamp-fips
いいえはいSep 02, 2026
CVE-2026-56855HIGH7.5
  • Docker logoDocker
  • argo-workflows-3.7
いいえはいSep 02, 2026
CVE-2026-75593HIGH7.2
  • Docker logoDocker
  • container-tools:rhel8::podman-gvproxy
いいえいいえAug 19, 2026
CVE-2026-61711MEDIUM5.3
  • Docker logoDocker
  • docker-compose
いいえはいAug 19, 2026
CVE-2026-61712LOW2.3
  • Docker logoDocker
  • conftest-fips
いいえはいAug 19, 2026

無料の脆弱性評価

クラウドセキュリティポスチャーのベンチマーク

9つのセキュリティドメインにわたるクラウドセキュリティプラクティスを評価して、リスクレベルをベンチマークし、防御のギャップを特定します。

評価を依頼する

パーソナライズされたデモを見る

実際に Wiz を見てみませんか?​

"私が今まで見た中で最高のユーザーエクスペリエンスは、クラウドワークロードを完全に可視化します。"
デビッド・エストリックCISO (最高情報責任者)
"Wiz を使えば、クラウド環境で何が起こっているかを 1 つの画面で確認することができます"
アダム・フレッチャーチーフ・セキュリティ・オフィサー
"Wizが何かを重要視した場合、それは実際に重要であることを私たちは知っています。"
グレッグ・ポニャトフスキ脅威および脆弱性管理責任者