[lit] Adjust error handling for decode introduced by r374665

On that decode, Windows bots fail with:

```
UnicodeEncodeError: 'ascii' codec can't encode characters in position 7-8: ordinal not in range(128)
```

That's the same error as before r374665 except it's now at the decode
before the write to stdout.

llvm-svn: 374666

1 files changed, 1 insertions, 1 deletions

diff --git a/llvm/utils/lit/lit/builtin_commands/diff.py b/llvm/utils/lit/lit/builtin_commands/diff.py
index ee933d8e59e..d851c4cb70d 100644
--- a/llvm/utils/lit/lit/builtin_commands/diff.py
+++ b/llvm/utils/lit/lit/builtin_commands/diff.py
@@ -97,7 +97,7 @@ def compareTwoTextFiles(flags, filepaths, filelines_bin, encoding):
                      n = flags.num_context_lines):
         if hasattr(diff, 'decode'):
             # python 2.7
-            diff = diff.decode()
+            diff = diff.decode(errors="backslashreplace")
         sys.stdout.write(diff)
         exitCode = 1
     return exitCode