On Wed, Feb 22, 2017 at 4:51 AM, Evan Lloyd Evan.Lloyd@arm.com wrote:
I doubt that gcc is messing with file names. I would suspect a newlib change.
So does that imply the GCC behaviour change will not get reverted?
I haven't seen a GCC behaviour change yet. I got the mingw32 gcc-linaro-5.3-2016.05 working. When I run gcc -E on a TMP.C file that includes a TMP.H file, I see both TMP.C ahd TMP.H in the -E output as expected. I see the same behaviour from a mingw gcc-linaro-4.9-2016.02. Unfortunately, I can't run any linaro release I've tried from 2016.08 or later and I don't know why. I don't know what you might be doing different that might be generating different results.
I looked at the gcc and newlib sources, but I don't see anything obvious downcasing filenames except for some filename hash table code in libiberty which doesn't appear to be used by either gcc or newlib.
Jim