Fix r357749 for big-endian architectures

We need to read the strings from the minidump files as little-endian,
regardless of the host byte order.

I definitely remember thinking about this case while writing the patch
(and in fact, I have implemented that for the "write" case), but somehow
I have ended up not implementing the byte swapping when reading the
data. This adds the necessary byte-swapping and should hopefully fix
test failures on big-endian bots.

llvm-svn: 357754
diff --git a/llvm/lib/Object/Minidump.cpp b/llvm/lib/Object/Minidump.cpp
index 3e3255a..1a22491 100644
--- a/llvm/lib/Object/Minidump.cpp
+++ b/llvm/lib/Object/Minidump.cpp
@@ -38,12 +38,16 @@
     return "";
 
   Offset += sizeof(support::ulittle32_t);
-  auto ExpectedData = getDataSliceAs<UTF16>(getData(), Offset, Size);
+  auto ExpectedData =
+      getDataSliceAs<support::ulittle16_t>(getData(), Offset, Size);
   if (!ExpectedData)
     return ExpectedData.takeError();
 
+  SmallVector<UTF16, 32> WStr(Size);
+  copy(*ExpectedData, WStr.begin());
+
   std::string Result;
-  if (!convertUTF16ToUTF8String(*ExpectedData, Result))
+  if (!convertUTF16ToUTF8String(WStr, Result))
     return createError("String decoding failed");
 
   return Result;