Merged bug25860 into default

3 files changed, 32 insertions, 3 deletions

diff --git a/codegen.py b/codegen.py
index 842549cf..91fa1154 100644
--- a/codegen.py
+++ b/codegen.py
@@ -187,7 +187,7 @@ def genErl(spec):
         elif type == 'table':
             return p+'Len:32/unsigned, '+p+'Tab:'+p+'Len/binary'
 
-    def genFieldPostprocessing(packed):
+    def genFieldPostprocessing(packed, hasContent):
         for f in packed:
             type = erlType(f.domain)
             if type == 'bit':
@@ -199,6 +199,10 @@ def genErl(spec):
             elif type == 'table':
                 print "  F%d = rabbit_binary_parser:parse_table(F%dTab)," % \
                       (f.index, f.index)
+            # We skip the check on content-bearing methods for
+            # speed. This is a sanity check, not a security thing.
+            elif type == 'shortstr' and not hasContent:
+                print "  rabbit_binary_parser:assert_utf8(F%d)," % (f.index)
             else:
                 pass
 
@@ -214,7 +218,7 @@ def genErl(spec):
             restSeparator = ''
         recordConstructorExpr = '#%s{%s}' % (m.erlangName(), fieldMapList(m.arguments))
         print "decode_method_fields(%s, <<%s>>) ->" % (m.erlangName(), binaryPattern)
-        genFieldPostprocessing(packedFields)
+        genFieldPostprocessing(packedFields, m.hasContent)
         print "  %s;" % (recordConstructorExpr,)
 
     def genDecodeProperties(c):
diff --git a/src/rabbit_binary_parser.erl b/src/rabbit_binary_parser.erl
index dc6d090f..088ad0e5 100644
--- a/src/rabbit_binary_parser.erl
+++ b/src/rabbit_binary_parser.erl
@@ -20,6 +20,7 @@
 
 -export([parse_table/1]).
 -export([ensure_content_decoded/1, clear_decoded_content/1]).
+-export([validate_utf8/1, assert_utf8/1]).
 
 %%----------------------------------------------------------------------------
 
@@ -30,6 +31,8 @@
         (rabbit_types:content()) -> rabbit_types:decoded_content()).
 -spec(clear_decoded_content/1 ::
         (rabbit_types:content()) -> rabbit_types:undecoded_content()).
+-spec(validate_utf8/1 :: (binary()) -> 'ok' | 'error').
+-spec(assert_utf8/1 :: (binary()) -> 'ok').
 
 -endif.
 
@@ -99,3 +102,18 @@ clear_decoded_content(Content = #content{properties_bin = none}) ->
     Content;
 clear_decoded_content(Content = #content{}) ->
     Content#content{properties = none}.
+
+assert_utf8(B) ->
+    case validate_utf8(B) of
+        ok    -> ok;
+        error -> rabbit_misc:protocol_error(
+                   frame_error, "Malformed UTF-8 in shortstr", [])
+    end.
+
+validate_utf8(Bin) ->
+    try
+        xmerl_ucs:from_utf8(Bin),
+        ok
+    catch exit:{ucs, _} ->
+            error
+    end.
diff --git a/src/rabbit_control_main.erl b/src/rabbit_control_main.erl
index 6f36f99d..f3463286 100644
--- a/src/rabbit_control_main.erl
+++ b/src/rabbit_control_main.erl
@@ -706,7 +706,14 @@ unsafe_rpc(Node, Mod, Fun, Args) ->
     end.
 
 call(Node, {Mod, Fun, Args}) ->
-    rpc_call(Node, Mod, Fun, lists:map(fun list_to_binary/1, Args)).
+    rpc_call(Node, Mod, Fun, lists:map(fun list_to_binary_utf8/1, Args)).
+
+list_to_binary_utf8(L) ->
+    B = list_to_binary(L),
+    case rabbit_binary_parser:validate_utf8(B) of
+        ok    -> B;
+        error -> throw({error, {not_utf_8, L}})
+    end.
 
 rpc_call(Node, Mod, Fun, Args) ->
     rpc:call(Node, Mod, Fun, Args, ?RPC_TIMEOUT).