CVE-2026-42858

@@ -97,7 +97,13 @@ def fetch_saml_metadata():
                     num_updated += 1
                 else:
                     log.info(f"→ Updated existing SAMLProviderData. Nothing has changed for entityID {entity_id}")
-        except (exceptions.SSLError, exceptions.HTTPError, exceptions.RequestException, MetadataParseError, SAMLMetadataURLError) as error:
+        except (
+            exceptions.SSLError,
+            exceptions.HTTPError,
+            exceptions.RequestException,
+            MetadataParseError,
+            SAMLMetadataURLError,
+        ) as error:
             # Catch and process exception in case of errors during fetching and processing saml metadata.
             # Here is a description of each exception.
             # SSLError is raised in case of errors caused by SSL (e.g. SSL cer verification failure etc.)

@@ -1925,16 +1925,19 @@ def add_optional_apps(optional_apps, installed_apps):
 SOCIAL_AUTH_SAML_SP_PRIVATE_KEY_DICT = {}
 SOCIAL_AUTH_SAML_SP_PUBLIC_CERT_DICT = {}
 
-# .. setting_name: SAML_METADATA_URL_ALLOW_PRIVATE_IPS
-# .. setting_default: False
-# .. setting_description: When False (the default), fetching SAML metadata from
+# .. toggle_name: SAML_METADATA_URL_ALLOW_PRIVATE_IPS
+# .. toggle_default: False
+# .. toggle_description: When False (the default), fetching SAML metadata from
 #   private IP address ranges (RFC 1918: 10.x, 172.16.x, 192.168.x) is blocked
 #   as a defense against SSRF attacks. Set to True only in deployments where the
 #   SAML Identity Provider is hosted on the same private network as the Open edX
 #   server. Note: loopback (127.x) and link-local (169.254.x) addresses remain
-#   blocked regardless of this setting. Operators are also encouraged to enforce
+#   blocked regardless of this toggle. Operators are also encouraged to enforce
 #   network-level egress filtering as a complementary control, particularly to
 #   cover hostname-based URLs that are not subject to IP validation.
+# .. toggle_implementation: DjangoSetting
+# .. toggle_use_cases: open_edx
+# .. toggle_creation_date: 2026-04-24
 SAML_METADATA_URL_ALLOW_PRIVATE_IPS = False
 
 ########################### django-fernet-fields ###########################

@@ -16,8 +16,10 @@
 from common.djangoapps.third_party_auth.models import SAMLConfiguration, SAMLProviderConfig
 from common.djangoapps.third_party_auth.utils import (
     MetadataParseError,
+    SAMLMetadataURLError,
     create_or_update_bulk_saml_provider_data,
     parse_metadata_xml,
+    validate_saml_metadata_url,
 )
 
 log = logging.getLogger(__name__)
@@ -74,10 +76,9 @@ def fetch_saml_metadata():
     failure_messages = []  # We return the length of this array for num_failed
     for url, entity_ids in url_map.items():
         try:
+            validate_saml_metadata_url(url)
             log.info("Fetching %s", url)
-            if not url.lower().startswith('https'):
-                log.warning("This SAML metadata URL is not secure! It should use HTTPS. (%s)", url)
-            response = requests.get(url, verify=True)  # May raise HTTPError or SSLError or ConnectionError
+            response = requests.get(url, verify=True, timeout=30)  # May raise HTTPError or SSLError or ConnectionError
             response.raise_for_status()  # May raise an HTTPError
 
             try:
@@ -96,7 +97,7 @@ def fetch_saml_metadata():
                     num_updated += 1
                 else:
                     log.info(f"→ Updated existing SAMLProviderData. Nothing has changed for entityID {entity_id}")
-        except (exceptions.SSLError, exceptions.HTTPError, exceptions.RequestException, MetadataParseError) as error:
+        except (exceptions.SSLError, exceptions.HTTPError, exceptions.RequestException, MetadataParseError, SAMLMetadataURLError) as error:
             # Catch and process exception in case of errors during fetching and processing saml metadata.
             # Here is a description of each exception.
             # SSLError is raised in case of errors caused by SSL (e.g. SSL cer verification failure etc.)

@@ -6,19 +6,23 @@
 from unittest.mock import MagicMock
 
 import ddt
+import pytest
+from django.test import override_settings
 from lxml import etree
 
 from common.djangoapps.student.tests.factories import UserFactory
 from common.djangoapps.third_party_auth.models import SAMLProviderData
 from common.djangoapps.third_party_auth.tests.testutil import TestCase
 from common.djangoapps.third_party_auth.utils import (
+    SAMLMetadataURLError,
     create_or_update_bulk_saml_provider_data,
     get_associated_user_by_email_response,
     get_user_from_email,
     is_enterprise_customer_user,
     is_oauth_provider,
     parse_metadata_xml,
     user_exists,
+    validate_saml_metadata_url,
 )
 from openedx.core.djangolib.testing.utils import skip_unless_lms
 from openedx.features.enterprise_support.tests.factories import (
@@ -272,3 +276,60 @@ def test_multiple_keys(self):
             entity_id='http://entity1',
         ).count()
         assert count == 2
+
+
+@ddt.ddt
+@skip_unless_lms
+class TestValidateSAMLMetadataURL(TestCase):
+    """Tests for validate_saml_metadata_url."""
+
+    @ddt.data(
+        'https://idp.example.com/metadata',
+        'https://1.1.1.1/metadata',
+    )
+    def test_valid_urls_pass(self, url):
+        validate_saml_metadata_url(url)  # should not raise
+
+    @ddt.data(
+        ('http://idp.example.com/metadata', 'must use HTTPS'),
+        ('ftp://idp.example.com/metadata', 'must use HTTPS'),
+        ('https://', 'no hostname'),
+    )
+    @ddt.unpack
+    def test_invalid_scheme_or_missing_hostname(self, url, expected_fragment):
+        with pytest.raises(SAMLMetadataURLError, match=expected_fragment):
+            validate_saml_metadata_url(url)
+
+    @ddt.data(
+        'https://127.0.0.1/metadata',       # IPv4 loopback
+        'https://[::1]/metadata',            # IPv6 loopback
+        'https://169.254.169.254/latest',    # AWS metadata endpoint
+        'https://169.254.0.1/metadata',      # other link-local
+        'https://[fe80::1]/metadata',        # IPv6 link-local
+        'https://240.0.0.1/metadata',        # reserved (Class E)
+    )
+    def test_always_blocked_regardless_of_setting(self, url):
+        for allow_private in (False, True):
+            with override_settings(SAML_METADATA_URL_ALLOW_PRIVATE_IPS=allow_private):
+                with pytest.raises(SAMLMetadataURLError):
+                    validate_saml_metadata_url(url)
+
+    @ddt.data(
+        'https://10.0.0.1/metadata',
+        'https://172.16.0.1/metadata',
+        'https://192.168.1.1/metadata',
+        'https://[fc00::1]/metadata',        # IPv6 unique local
+    )
+    def test_private_ip_blocked_by_default(self, url):
+        with pytest.raises(SAMLMetadataURLError):
+            validate_saml_metadata_url(url)
+
+    @ddt.data(
+        'https://10.0.0.1/metadata',
+        'https://172.16.0.1/metadata',
+        'https://192.168.1.1/metadata',
+        'https://[fc00::1]/metadata',        # IPv6 unique local
+    )
+    @override_settings(SAML_METADATA_URL_ALLOW_PRIVATE_IPS=True)
+    def test_private_ip_allowed_when_setting_enabled(self, url):
+        validate_saml_metadata_url(url)  # should not raise

@@ -3,9 +3,12 @@
 """
 
 import datetime
+import ipaddress
+from urllib.parse import urlparse
 from zoneinfo import ZoneInfo
 
 import dateutil.parser
+from django.conf import settings
 from django.contrib.auth.models import User  # lint-amnesty, pylint: disable=imported-auth-user
 from django.utils.timezone import now
 from enterprise.models import EnterpriseCustomerIdentityProvider, EnterpriseCustomerUser
@@ -27,6 +30,67 @@ class MetadataParseError(Exception):
     pass  # lint-amnesty, pylint: disable=unnecessary-pass
 
 
+class SAMLMetadataURLError(Exception):
+    """ A SAML metadata URL failed security validation """
+    pass  # lint-amnesty, pylint: disable=unnecessary-pass
+
+
+def validate_saml_metadata_url(url):
+    """
+    Validate that a SAML metadata URL is safe to fetch.
+
+    Enforces HTTPS and blocks requests to loopback, link-local, and reserved IP
+    addresses. Link-local specifically covers cloud instance metadata endpoints
+    (169.254.0.0/16, e.g. the AWS metadata service at 169.254.169.254).
+    Reserved addresses (e.g. 240.0.0.0/4) are IETF-assigned ranges that are
+    never routable on real networks.
+
+    Private IP ranges (RFC 1918: 10.x, 172.16.x, 192.168.x) are also blocked by
+    default, since most Open edX deployments fetch SAML metadata from public IdPs.
+    Operators running in a private network where the SAML IdP has a private IP can
+    opt out by setting SAML_METADATA_URL_ALLOW_PRIVATE_IPS = True in Django settings.
+
+    Limitation: IP address checks only apply to literal IPs in the URL. Hostname-
+    based URLs are not validated against the IP blocklists. Operators are encouraged
+    to complement this with network-level egress filtering that blocks outbound
+    connections from the Open edX server to link-local (169.254.0.0/16) and RFC
+    1918 private address ranges.
+
+    Raises SAMLMetadataURLError if the URL fails validation.
+    """
+    parsed = urlparse(url)
+
+    if parsed.scheme != 'https':
+        raise SAMLMetadataURLError(
+            f"SAML metadata URL must use HTTPS, got scheme: {parsed.scheme!r}"
+        )
+
+    hostname = parsed.hostname
+    if not hostname:
+        raise SAMLMetadataURLError("SAML metadata URL has no hostname")
+
+    try:
+        addr = ipaddress.ip_address(hostname)
+    except ValueError:
+        # hostname is a domain name, not a numeric IP literal — pass through.
+        return
+
+    # Loopback, link-local, and reserved ranges are never legitimate SAML IdP
+    # addresses regardless of deployment topology.
+    if addr.is_loopback or addr.is_link_local or addr.is_reserved:
+        raise SAMLMetadataURLError(
+            f"SAML metadata URL hostname is a forbidden IP address: {addr}"
+        )
+
+    # Private ranges are blocked by default but can be allowed via Django settings
+    # for deployments where the SAML IdP lives on the same private network.
+    if addr.is_private and not settings.SAML_METADATA_URL_ALLOW_PRIVATE_IPS:
+        raise SAMLMetadataURLError(
+            f"SAML metadata URL hostname is a private IP address: {addr}. "
+            "Set SAML_METADATA_URL_ALLOW_PRIVATE_IPS = True in Django settings to allow this."
+        )
+
+
 def parse_metadata_xml(xml, entity_id):
     """
     Given an XML document containing SAML 2.0 metadata, parse it and return a tuple of

@@ -1925,6 +1925,18 @@ def add_optional_apps(optional_apps, installed_apps):
 SOCIAL_AUTH_SAML_SP_PRIVATE_KEY_DICT = {}
 SOCIAL_AUTH_SAML_SP_PUBLIC_CERT_DICT = {}
 
+# .. setting_name: SAML_METADATA_URL_ALLOW_PRIVATE_IPS
+# .. setting_default: False
+# .. setting_description: When False (the default), fetching SAML metadata from
+#   private IP address ranges (RFC 1918: 10.x, 172.16.x, 192.168.x) is blocked
+#   as a defense against SSRF attacks. Set to True only in deployments where the
+#   SAML Identity Provider is hosted on the same private network as the Open edX
+#   server. Note: loopback (127.x) and link-local (169.254.x) addresses remain
+#   blocked regardless of this setting. Operators are also encouraged to enforce
+#   network-level egress filtering as a complementary control, particularly to
+#   cover hostname-based URLs that are not subject to IP validation.
+SAML_METADATA_URL_ALLOW_PRIVATE_IPS = False
+
 ########################### django-fernet-fields ###########################
 
 FERNET_KEYS = [