Work around nytimes.com's broken hostnames in our SOCKS checks.

RFC 952 is approximately 30 years old, and people are failing to comply,
by serving A records with '_' as part of the hostname.  Since relaxing
the check is a QOL improvement for our userbase, relax the check to
allow such abominations as destinations, especially since there are
likely to be other similarly misconfigured domains out there.

diff --git a/changes/bug16430 b/changes/bug16430
new file mode 100644 (file)
index 0000000..ca7b874
--- /dev/null
+++ b/changes/bug16430
@@ -0,0 +1,4 @@
+  o Minor features (client):
+    - Relax the validation done to hostnames in SOCKS5 requests, and allow
+      '_' to cope with domains observed in the wild that are serving non-RFC
+      compliant records. Resolves ticket 16430.

diff --git a/src/common/util.c b/src/common/util.c
index 942d0c290e52813d74470d1e44c1e115572123ba..449015054fe445b1c85b2444aa85afff9b1b6e8e 100644 (file)
--- a/src/common/util.c
+++ b/src/common/util.c
@@ -1036,6 +1036,9 @@ string_is_valid_ipv6_address(const char *string)
 
 /** Return true iff <b>string</b> matches a pattern of DNS names
  * that we allow Tor clients to connect to.
+ *
+ * Note: This allows certain technically invalid characters ('_') to cope
+ * with misconfigured zones that have been encountered in the wild.
  */
 int
 string_is_valid_hostname(const char *string)
@@ -1048,7 +1051,7 @@ string_is_valid_hostname(const char *string)
   smartlist_split_string(components,string,".",0,0);
 
   SMARTLIST_FOREACH_BEGIN(components, char *, c) {
-    if (c[0] == '-') {
+    if ((c[0] == '-') || (*c == '_')) {
       result = 0;
       break;
     }
@@ -1057,7 +1060,7 @@ string_is_valid_hostname(const char *string)
       if ((*c >= 'a' && *c <= 'z') ||
           (*c >= 'A' && *c <= 'Z') ||
           (*c >= '0' && *c <= '9') ||
-          (*c == '-'))
+          (*c == '-') || (*c == '_'))
         c++;
       else
         result = 0;

diff --git a/src/test/test_util.c b/src/test/test_util.c
index b0366db37f75e5031f8544a6ef3265f9b2b81e8a..0f64c26e018e84275c1e355fe4976adbbeba232b 100644 (file)
--- a/src/test/test_util.c
+++ b/src/test/test_util.c
@@ -4268,18 +4268,23 @@ test_util_hostname_validation(void *arg)
   tt_assert(string_is_valid_hostname("stanford.edu"));
   tt_assert(string_is_valid_hostname("multiple-words-with-hypens.jp"));
 
-  // Subdomain name cannot start with '-'.
+  // Subdomain name cannot start with '-' or '_'.
   tt_assert(!string_is_valid_hostname("-torproject.org"));
   tt_assert(!string_is_valid_hostname("subdomain.-domain.org"));
   tt_assert(!string_is_valid_hostname("-subdomain.domain.org"));
+  tt_assert(!string_is_valid_hostname("___abc.org"));
 
   // Hostnames cannot contain non-alphanumeric characters.
   tt_assert(!string_is_valid_hostname("%%domain.\\org."));
   tt_assert(!string_is_valid_hostname("***x.net"));
-  tt_assert(!string_is_valid_hostname("___abc.org"));
   tt_assert(!string_is_valid_hostname("\xff\xffxyz.org"));
   tt_assert(!string_is_valid_hostname("word1 word2.net"));
 
+  // Test workaround for nytimes.com stupidity, technically invalid,
+  // but we allow it since they are big, even though they are failing to
+  // comply with a ~30 year old standard.
+  tt_assert(string_is_valid_hostname("core3_euw1.fabrik.nytimes.com"));
+
   // XXX: do we allow single-label DNS names?
 
   done: