CVE-2023-1786.patch

From 9d4903bacda67db0e65c7e0bd68c2efc90feab5b Mon Sep 17 00:00:00 2001
From: James Falcon <james.falcon@canonical.com>
Date: Mon, 3 Apr 2023 16:46:43 -0500
Subject: [PATCH] Make user/vendor data sensitive and remove log permissions

Because user data and vendor data may contain sensitive information,
this commit ensures that any user data or vendor data written to
instance-data.json gets redacted and is only available to root user.

Also, modify the permissions of cloud-init.log to be 640, so that
sensitive data leaked to the log isn't world readable.
Additionally, remove the logging of user data and vendor data to
cloud-init.log from the Vultr datasource.

LP: #2013967
CVE: CVE-2023-1786
---
 cloudinit/sources/DataSourceLXD.py   |  9 ++++++---
 cloudinit/sources/DataSourceVultr.py | 14 ++++++--------
 cloudinit/sources/__init__.py        | 28 +++++++++++++++++++++++++---
 cloudinit/stages.py                  |  4 +++-
 tests/unittests/sources/test_init.py | 27 ++++++++++++++++++++++++++-
 tests/unittests/test_stages.py       | 18 +++++++++++-------
 6 files changed, 77 insertions(+), 23 deletions(-)

diff --git a/cloudinit/sources/DataSourceLXD.py b/cloudinit/sources/DataSourceLXD.py
index ab440cc8d0..e4cae91abf 100644
--- a/cloudinit/sources/DataSourceLXD.py
+++ b/cloudinit/sources/DataSourceLXD.py
@@ -14,7 +14,7 @@
 import time
 from enum import Flag, auto
 from json.decoder import JSONDecodeError
-from typing import Any, Dict, List, Optional, Union, cast
+from typing import Any, Dict, List, Optional, Tuple, Union, cast
 
 import requests
 from requests.adapters import HTTPAdapter
@@ -168,11 +168,14 @@ class DataSourceLXD(sources.DataSource):
     _network_config: Union[Dict, str] = sources.UNSET
     _crawled_metadata: Union[Dict, str] = sources.UNSET
 
-    sensitive_metadata_keys = (
-        "merged_cfg",
+    sensitive_metadata_keys: Tuple[
+        str, ...
+    ] = sources.DataSource.sensitive_metadata_keys + (
         "user.meta-data",
         "user.vendor-data",
         "user.user-data",
+        "cloud-init.user-data",
+        "cloud-init.vendor-data",
     )
 
     skip_hotplug_detect = True
diff --git a/cloudinit/sources/DataSourceVultr.py b/cloudinit/sources/DataSourceVultr.py
index 9d7c84fbf0..660e9f142b 100644
--- a/cloudinit/sources/DataSourceVultr.py
+++ b/cloudinit/sources/DataSourceVultr.py
@@ -5,6 +5,8 @@
 # Vultr Metadata API:
 # https://www.vultr.com/metadata/
 
+from typing import Tuple
+
 import cloudinit.sources.helpers.vultr as vultr
 from cloudinit import log as log
 from cloudinit import sources, util, version
@@ -28,6 +30,10 @@ class DataSourceVultr(sources.DataSource):
 
     dsname = "Vultr"
 
+    sensitive_metadata_keys: Tuple[
+        str, ...
+    ] = sources.DataSource.sensitive_metadata_keys + ("startup-script",)
+
     def __init__(self, sys_cfg, distro, paths):
         super(DataSourceVultr, self).__init__(sys_cfg, distro, paths)
         self.ds_cfg = util.mergemanydict(
@@ -54,13 +60,8 @@ def _get_data(self):
         self.get_datasource_data(self.metadata)
 
         # Dump some data so diagnosing failures is manageable
-        LOG.debug("Vultr Vendor Config:")
-        LOG.debug(util.json_dumps(self.metadata["vendor-data"]))
         LOG.debug("SUBID: %s", self.metadata["instance-id"])
         LOG.debug("Hostname: %s", self.metadata["local-hostname"])
-        if self.userdata_raw is not None:
-            LOG.debug("User-Data:")
-            LOG.debug(self.userdata_raw)
 
         return True
 
@@ -146,7 +147,4 @@ def get_datasource_list(depends):
     config = md["vendor-data"]
     sysinfo = vultr.get_sysinfo()
 
-    print(util.json_dumps(sysinfo))
-    print(util.json_dumps(config))
-
 # vi: ts=4 expandtab
diff --git a/cloudinit/sources/__init__.py b/cloudinit/sources/__init__.py
index 565e17549b..5c6ae8b18c 100644
--- a/cloudinit/sources/__init__.py
+++ b/cloudinit/sources/__init__.py
@@ -110,7 +110,10 @@ def process_instance_metadata(metadata, key_path="", sensitive_keys=()):
             sub_key_path = key_path + "/" + key
         else:
             sub_key_path = key
-        if key in sensitive_keys or sub_key_path in sensitive_keys:
+        if (
+            key.lower() in sensitive_keys
+            or sub_key_path.lower() in sensitive_keys
+        ):
             sens_keys.append(sub_key_path)
         if isinstance(val, str) and val.startswith("ci-b64:"):
             base64_encoded_keys.append(sub_key_path)
@@ -132,6 +135,12 @@ def redact_sensitive_keys(metadata, redact_value=REDACT_SENSITIVE_VALUE):
 
     Replace any keys values listed in 'sensitive_keys' with redact_value.
     """
+    # While 'sensitive_keys' should already sanitized to only include what
+    # is in metadata, it is possible keys will overlap. For example, if
+    # "merged_cfg" and "merged_cfg/ds/userdata" both match, it's possible that
+    # "merged_cfg" will get replaced first, meaning "merged_cfg/ds/userdata"
+    # no longer represents a valid key.
+    # Thus, we still need to do membership checks in this function.
     if not metadata.get("sensitive_keys", []):
         return metadata
     md_copy = copy.deepcopy(metadata)
@@ -139,9 +148,14 @@ def redact_sensitive_keys(metadata, redact_value=REDACT_SENSITIVE_VALUE):
         path_parts = key_path.split("/")
         obj = md_copy
         for path in path_parts:
-            if isinstance(obj[path], dict) and path != path_parts[-1]:
+            if (
+                path in obj
+                and isinstance(obj[path], dict)
+                and path != path_parts[-1]
+            ):
                 obj = obj[path]
-        obj[path] = redact_value
+        if path in obj:
+            obj[path] = redact_value
     return md_copy
 
 
@@ -249,6 +263,14 @@ class DataSource(CloudInitPickleMixin, metaclass=abc.ABCMeta):
     sensitive_metadata_keys: Tuple[str, ...] = (
         "merged_cfg",
         "security-credentials",
+        "userdata",
+        "user-data",
+        "user_data",
+        "vendordata",
+        "vendor-data",
+        # Provide ds/vendor_data to avoid redacting top-level
+        #  "vendor_data": {enabled: True}
+        "ds/vendor_data",
     )
 
     # True on datasources that may not see hotplugged devices reflected
diff --git a/cloudinit/stages.py b/cloudinit/stages.py
index 9494a0bfa2..21f30a1f16 100644
--- a/cloudinit/stages.py
+++ b/cloudinit/stages.py
@@ -203,7 +203,9 @@ def _initialize_filesystem(self):
         util.ensure_dirs(self._initial_subdirs())
         log_file = util.get_cfg_option_str(self.cfg, "def_log_file")
         if log_file:
-            util.ensure_file(log_file, mode=0o640, preserve_mode=True)
+            # At this point the log file should have already been created
+            # in the setupLogging function of log.py
+            util.ensure_file(log_file, mode=0o640, preserve_mode=False)
             perms = self.cfg.get("syslog_fix_perms")
             if not perms:
                 perms = {}
diff --git a/tests/unittests/sources/test_init.py b/tests/unittests/sources/test_init.py
index 0447e02cb6..eb27198f9b 100644
--- a/tests/unittests/sources/test_init.py
+++ b/tests/unittests/sources/test_init.py
@@ -458,12 +458,24 @@ def test_get_data_writes_redacted_public_json_instance_data(self):
                         "cred2": "othersekret",
                     }
                 },
+                "someother": {
+                    "nested": {
+                        "userData": "HIDE ME",
+                    }
+                },
+                "VENDOR-DAta": "HIDE ME TOO",
             },
         )
         self.assertCountEqual(
             (
                 "merged_cfg",
                 "security-credentials",
+                "userdata",
+                "user-data",
+                "user_data",
+                "vendordata",
+                "vendor-data",
+                "ds/vendor_data",
             ),
             datasource.sensitive_metadata_keys,
         )
@@ -490,7 +502,9 @@ def test_get_data_writes_redacted_public_json_instance_data(self):
             "base64_encoded_keys": [],
             "merged_cfg": REDACT_SENSITIVE_VALUE,
             "sensitive_keys": [
+                "ds/meta_data/VENDOR-DAta",
                 "ds/meta_data/some/security-credentials",
+                "ds/meta_data/someother/nested/userData",
                 "merged_cfg",
             ],
             "sys_info": sys_info,
@@ -500,6 +514,7 @@ def test_get_data_writes_redacted_public_json_instance_data(self):
                 "availability_zone": "myaz",
                 "cloud-name": "subclasscloudname",
                 "cloud_name": "subclasscloudname",
+                "cloud_id": "subclasscloudname",
                 "distro": "ubuntu",
                 "distro_release": "focal",
                 "distro_version": "20.04",
@@ -522,14 +537,18 @@ def test_get_data_writes_redacted_public_json_instance_data(self):
             "ds": {
                 "_doc": EXPERIMENTAL_TEXT,
                 "meta_data": {
+                    "VENDOR-DAta": REDACT_SENSITIVE_VALUE,
                     "availability_zone": "myaz",
                     "local-hostname": "test-subclass-hostname",
                     "region": "myregion",
                     "some": {"security-credentials": REDACT_SENSITIVE_VALUE},
+                    "someother": {
+                        "nested": {"userData": REDACT_SENSITIVE_VALUE}
+                    },
                 },
             },
         }
-        self.assertCountEqual(expected, redacted)
+        self.assertEqual(expected, redacted)
         file_stat = os.stat(json_file)
         self.assertEqual(0o644, stat.S_IMODE(file_stat.st_mode))
 
@@ -574,6 +593,12 @@ def test_get_data_writes_json_instance_data_sensitive(self):
             (
                 "merged_cfg",
                 "security-credentials",
+                "userdata",
+                "user-data",
+                "user_data",
+                "vendordata",
+                "vendor-data",
+                "ds/vendor_data",
             ),
             datasource.sensitive_metadata_keys,
         )
diff --git a/tests/unittests/test_stages.py b/tests/unittests/test_stages.py
index 15a7e973dc..a61f9df9b8 100644
--- a/tests/unittests/test_stages.py
+++ b/tests/unittests/test_stages.py
@@ -606,19 +606,23 @@ def test_log_files_existence_is_ensured_if_configured(self, init, tmpdir):
         # Assert we create it 0o640  by default if it doesn't already exist
         assert 0o640 == stat.S_IMODE(log_file.stat().mode)
 
-    def test_existing_file_permissions_are_not_modified(self, init, tmpdir):
-        """If the log file already exists, we should not modify its permissions
+    def test_existing_file_permissions(self, init, tmpdir):
+        """Test file permissions are set as expected.
+
+        CIS Hardening requires 640 permissions. These permissions are
+        currently hardcoded on every boot, but if there's ever a reason
+        to change this, we need to then ensure that they
+        are *not* set every boot.
 
         See https://bugs.launchpad.net/cloud-init/+bug/1900837.
         """
-        # Use a mode that will never be made the default so this test will
-        # always be valid
-        mode = 0o606
         log_file = tmpdir.join("cloud-init.log")
         log_file.ensure()
-        log_file.chmod(mode)
+        # Use a mode that will never be made the default so this test will
+        # always be valid
+        log_file.chmod(0o606)
         init._cfg = {"def_log_file": str(log_file)}
 
         init._initialize_filesystem()
 
-        assert mode == stat.S_IMODE(log_file.stat().mode)
+        assert 0o640 == stat.S_IMODE(log_file.stat().mode)