"Or to put it differently, should we put a specific fix for a single
gcc version into memoryview.c? For all we know, a fix will be pushed
to Fedora 28 in the next 4 months."

Right now, test_buffer is skipped in the python3.spec of Fedora (recipe to build the package which compiles Python and then runs the Python test suite) on pp64le because of this bug. With my packager hat: I'm done, the bug has been identified and is already worked around in the package.

With my Python hat: I'm not sure if I'm comfortable to have a known rounding issue on memoryview because of a compiler bug. I'm not sure that the GCC bug will be fixed quickly, nor if all operating systems will quickly update GCC or backport the fix.

Well, I wrote a fix, but I don't feel able to decide if the workaround is worth it or not.

For me, the main risk is to forget to remove the workaround once a new GCC version will be released (in N months).

I wouldn't bet that such GCC bug will be fixed and released in less than 4 months.