https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65159
--- Comment #12 from David Krauss <potswa at mac dot com> --- I did exactly the steps that I mentioned: cleanly build r220825, update to r220000, and build again without cleaning first. The object being to roughly replicate the conditions when the bug was originally observed — except originally the update was in the other direction, r220000-ish to r220825. Is it always an error to do a non-clean build? The problem seems to happen pretty regularly.
