This patch helps infer the endianness of DWARF sections from FileHeader.
Same comments generally apply to this test case as the big endian one from D80972. Otherwise, this looks fine.
You're welcome to continue to use %t1.o and %t2.o etc if you want - my previous suggestion in this area was more by way of example. What you have here is fine too though.
%t.be.o and %t.le.o belong to the test case (a). I want to avoid having something like:
# RUN: yaml2obj --docnum=1 %s -o %t1.o ## Big endian case. ... # RUN: yaml2obj --docnum=1 %s -o %t2.o ## Little endian case. ... # RUN: yaml2obj --dconum=2 %s -o %t3.o