Fixes for re-enabling more MSVC level 4 warnings: media/ edition

media/cast/rtcp/test_rtcp_packet_builder.cc

Index: media/cast/rtcp/test_rtcp_packet_builder.cc
diff --git a/media/cast/rtcp/test_rtcp_packet_builder.cc b/media/cast/rtcp/test_rtcp_packet_builder.cc
index 5bfebf6cb9da254128700c4469dba7036569e1e5..fc1235bfbe52a472b73a869576c237f5658355ab 100644
--- a/media/cast/rtcp/test_rtcp_packet_builder.cc
+++ b/media/cast/rtcp/test_rtcp_packet_builder.cc
@@ -65,21 +65,7 @@ void TestRtcpPacketBuilder::AddSdesCname(uint32 sender_ssrc,
   for (size_t i = 0; i < c_name.size(); ++i) {
     big_endian_writer_.WriteU8(c_name.c_str()[i]);
   }
-  int padding;
-  switch (c_name.size() % 4) {
-    case 0:
-      padding = 2;
-      break;
-    case 1:
-      padding = 1;
-      break;
-    case 2:
-      padding = 4;
-      break;
-    case 3:
-      padding = 3;
-      break;
-  }
+  const int padding = (6 - (c_name.size() % 4)) % 4;

[ddorwin, 2014/07/08 18:17:50]

"case 2" was 4 and is now 0.

[Peter Kasting, 2014/07/08 18:39:14]

Good catch.  Fixed.

Although I'm not sure it actually matters -- I assume this code is just here to
pad things out so the next item written will have 4-byte alignment, in which
case writing 4 bytes of padding is the same as writing no padding (except for
wasting some space?).  And the trybots seemed to pass, so it seems like
unittests don't rely on the distinction.

However, this patch isn't supposed to change behavior, so fixing it is the right
thing anyway.

   for (int j = 0; j < padding; ++j) {
     big_endian_writer_.WriteU8(0);
   }