Re-enable DrMemory test for ICOImageDecoderTests.parseAndDecodeByteByByte.

third_party/WebKit/Source/platform/image-decoders/ico/ICOImageDecoder.cpp

Index: third_party/WebKit/Source/platform/image-decoders/ico/ICOImageDecoder.cpp
diff --git a/third_party/WebKit/Source/platform/image-decoders/ico/ICOImageDecoder.cpp b/third_party/WebKit/Source/platform/image-decoders/ico/ICOImageDecoder.cpp
index f8d39b99ad264cef00633c49a40f6643b3af1b64..257caac7beb16399c5efd873ba3ccf27283b2572 100644
--- a/third_party/WebKit/Source/platform/image-decoders/ico/ICOImageDecoder.cpp
+++ b/third_party/WebKit/Source/platform/image-decoders/ico/ICOImageDecoder.cpp
@@ -162,9 +162,9 @@ void ICOImageDecoder::decode(size_t index, bool onlySize)
     // may have called SharedBuffer::mergeSegmentsIntoBuffer().
     m_fastReader.clearCache();
 
-    // If we couldn't decode the image but we've received all the data, decoding
-    // has failed.
-    if ((!decodeDirectory() || (!onlySize && !decodeAtIndex(index))) && isAllDataReceived()) {
+    // If we couldn't decode the image or the data was truncated but we've
+    // received all the data, decoding has failed.
+    if ((!decodeDirectory() || m_data->size() < m_decodedOffset || (!onlySize && !decodeAtIndex(index))) && isAllDataReceived()) {

[Peter Kasting, 2016/06/20 23:08:03]

Why do we want this behavior?

If we're not trying to actually decode the missing data, it seems like decoding
shouldn't fail until we do.

[aleksandar.stojiljkovic, 2016/06/20 23:30:24]

I have put explanation above, in decodeFrameCount where decodeSize is called.

Related to decoding frames, agree - that's why the check on truncation here is
only about header space at the beginning of the file (m_decodedOffset) and not
checking frames data truncation.

         setFailed();
     // If we're done decoding this frame, we don't need the BMPImageReader or
     // PNGImageDecoder anymore.  (If we failed, these have already been