Miscellaneous changes split from https://codereview.webrtc.org/1230503003 .

webrtc/modules/audio_coding/codecs/isac/main/source/isac.c

Index: webrtc/modules/audio_coding/codecs/isac/main/source/isac.c
diff --git a/webrtc/modules/audio_coding/codecs/isac/main/source/isac.c b/webrtc/modules/audio_coding/codecs/isac/main/source/isac.c
index a19fd01167c26e24b445b6d56e7f3021b3612d88..b410efc0473eab0ec7cbaf4a6964f6a62f85810f 100644
--- a/webrtc/modules/audio_coding/codecs/isac/main/source/isac.c
+++ b/webrtc/modules/audio_coding/codecs/isac/main/source/isac.c
@@ -750,7 +750,8 @@ int WebRtcIsac_Encode(ISACStruct* ISAC_main_inst,
                       streamLenUB + garbageLen, &crc);
 #ifndef WEBRTC_ARCH_BIG_ENDIAN
     for (k = 0; k < LEN_CHECK_SUM_WORD8; k++) {
-      encoded[streamLen - LEN_CHECK_SUM_WORD8 + k] = crc >> (24 - k * 8);
+      encoded[streamLen - LEN_CHECK_SUM_WORD8 + k] =
+          (crc >> (24 - k * 8)) & 0xFF;

[kwiberg-webrtc, 2015/07/12 18:40:30]

Why the explicit & 0xff?

[Peter Kasting, 2015/07/13 02:41:24]

It should probably be a static_cast<uint8_t>() instead, now that I think about
it.

The goal was to eliminate MSVC "implicit cast could have truncated value"
warnings by making the truncation explicit, but doing that with a cast is better
than doing it with a mask.

[Peter Kasting, 2015/07/13 22:25:04]

Changed to a static_cast.

[kwiberg-webrtc, 2015/07/15 01:10:53]

Excellent. (That you changed to a cast. A static_cast was maybe not such a good
idea... ;-) )

     }
 #else
     memcpy(&encoded[streamLenLB + streamLenUB + 1], &crc, LEN_CHECK_SUM_WORD8);