https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91947
Jonathan Wakely <redi at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|RESOLVED |ASSIGNED
Resolution|FIXED |---
--- Comment #12 from Jonathan Wakely <redi at gcc dot gnu.org> ---
Rather than a race, I think what might happen is that an interrupted build
leaves a partial file in place. If make is re-run it will think the file is
present and won't try to create that target.
So writing to a temporary file and then moving it avoids that. I don't think
the stamp file is strictly necessary.
I still don't see why the problem has appeared now, when we've been doing the
same thing for years for the other generated header.
Reopening ...
