MONGOCRYPT-960 further validate GCP fields

kms-message/src/kms_gcp_request.c

@@ -20,6 +20,8 @@
 #include "kms_message_private.h"
 #include "kms_request_opt_private.h"
 
+#include <ctype.h>
+
 /* Set a default expiration of 5 minutes for JSON Web Tokens (GCP allows up to
  * one hour) */
 #define JWT_EXPIRATION_SECS 5 * 60
@@ -145,6 +147,38 @@ kms_gcp_request_oauth_new (const char *host,
    return req;
 }
 
+/* _check_key_identifier returns true if `value` is safe to use as a URL path segment.
+ *
+ * Key identifiers may originate from key vault documents or caller-supplied options. They are
+ * formatted into the request path, so an unescaped '/', '?', '#', space, or CR/LF would rewrite
+ * the request target or inject header lines into the request.
+ *
+ * Only the RFC 3986 unreserved characters are permitted. This is a superset of what GCP itself
+ * accepts, so it does not reject any identifier that would otherwise work:
+ * - keyRing and keyName must match `[a-zA-Z0-9_-]{1,63}`. See
+ *   https://cloud.google.com/kms/docs/reference/rest/v1/projects.locations.keyRings/create and
+ *   https://cloud.google.com/kms/docs/reference/rest/v1/projects.locations.keyRings.cryptoKeys/create
+ * - projectId is 6-30 characters of lowercase letters, numbers, and hyphens. See
+ *   https://cloud.google.com/resource-manager/docs/creating-managing-projects
+ * - location is a Cloud location name (e.g. "global", "us-east1").
+ * - keyVersion is a positive integer.
+ *
+ * `value` is not included in the error message: it is untrusted, and the error may be logged. */
+static bool
+_check_key_identifier (kms_request_t *req, const char *name, const char *value)
+{
+   const char *c;
+
+   for (c = value; *c != '\0'; c++) {
+      if (!isalnum ((unsigned char) *c) && *c != '-' && *c != '.' && *c != '_' &&
+          *c != '~') {
+         KMS_ERROR (req, "Invalid character in GCP KMS key identifier: %s", name);
+         return false;
+      }
+   }
+   return true;
+}
+
 static kms_request_t *
 _encrypt_decrypt_common (const char *encrypt_decrypt,
                          const char *host,
@@ -192,6 +226,14 @@ _encrypt_decrypt_common (const char *encrypt_decrypt,
       goto done;
    }
 
+   if (!_check_key_identifier (req, "projectId", project_id) ||
+       !_check_key_identifier (req, "location", location) ||
+       !_check_key_identifier (req, "keyRing", key_ring_name) ||
+       !_check_key_identifier (req, "keyName", key_name) ||
+       (key_version && !_check_key_identifier (req, "keyVersion", key_version))) {
+      goto done;
+   }
+
    value_base64 = kms_message_raw_to_b64 (value, value_len);
    if (!value_base64) {
       KMS_ERROR (req, "Could not bases64-encode plaintext");

test/test-gcp-auth.c

@@ -242,7 +242,58 @@ static void _test_encrypt_with_accesstoken(_mongocrypt_tester_t *tester) {
     mongocrypt_destroy(crypt);
 }
 
+// Test a CRLF in a key vault document's keyName. Regression test for MONGOCRYPT-960.
+static void _test_crlf_in_gcp_masterkey(_mongocrypt_tester_t *tester) {
+    mongocrypt_t *crypt;
+    mongocrypt_ctx_t *ctx;
+    mongocrypt_binary_t *uuid;
+    const char *uuid_data = "\x61\x61\x61\x61\x61\x61\x61\x61\x61\x61\x61\x61\x61\x61\x61\x61";
+
+    // A key vault document whose masterKey.keyName contains a CRLF.
+    // clang-format off
+    const char *key_doc = BSON_STR({
+        "status" : 1,
+        "_id" : {"$binary" : {"base64" : "YWFhYWFhYWFhYWFhYWFhYQ==", "subType" : "04"}},
+        "masterKey" : {
+            "provider" : "gcp",
+            "projectId" : "test",
+            "location" : "global",
+            "keyRing" : "test",
+            "keyName" : "test\r\nFOOBAR: injected"
+        },
+        "updateDate" : {"$date" : {"$numberLong" : "1557827033449"}},
+        "creationDate" : {"$date" : {"$numberLong" : "1557827033449"}},
+        "keyMaterial" : {
+            "$binary" : {
+                "base64" : "PwwOESbKs57YTJtGSCsuAJbv9VWRHjLdPdziUVxH0K9woZUka4SghSwJlw9n6TU/dYHFODvSa4p7bfKzS6U8kGFvOnd7LcPU63IkTek4qZEbGpTRwI5lMT4FFWdpYpUf",
+                "subType" : "00"
+            }
+        }
+    });
+    // clang-format on
+
+    crypt = mongocrypt_new();
+    ASSERT_OK(mongocrypt_setopt_kms_providers(crypt, TEST_BSON("{'gcp': {'accessToken': 'foobar'}}")), crypt);
+    ASSERT_OK(_mongocrypt_init_for_test(crypt), crypt);
+    ctx = mongocrypt_ctx_new(crypt);
+    uuid = mongocrypt_binary_new_from_data((uint8_t *)uuid_data, UUID_LEN);
+    ASSERT_OK(mongocrypt_ctx_setopt_key_id(ctx, uuid), ctx);
+    ASSERT_OK(mongocrypt_ctx_setopt_algorithm(ctx, MONGOCRYPT_ALGORITHM_DETERMINISTIC_STR, -1), ctx);
+    ASSERT_OK(mongocrypt_ctx_explicit_encrypt_init(ctx, TEST_BSON("{'v': 1}")), ctx);
+
+    ASSERT_STATE_EQUAL(mongocrypt_ctx_state(ctx), MONGOCRYPT_CTX_NEED_MONGO_KEYS);
+    /* Constructing the KMS request fails. No request is sent. */
+    ASSERT_FAILS(mongocrypt_ctx_mongo_feed(ctx, TEST_BSON_STR(key_doc)),
+                 ctx,
+                 "Invalid character in GCP KMS key identifier: keyName");
+
+    mongocrypt_binary_destroy(uuid);
+    mongocrypt_ctx_destroy(ctx);
+    mongocrypt_destroy(crypt);
+}
+
 void _mongocrypt_tester_install_gcp_auth(_mongocrypt_tester_t *tester) {
+    INSTALL_TEST(_test_crlf_in_gcp_masterkey);
     INSTALL_TEST(_test_createdatakey_with_credentials);
     INSTALL_TEST(_test_encrypt_with_credentials);
     INSTALL_TEST(_test_createdatakey_with_accesstoken);