https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124001
--- Comment #5 from Joost VandeVondele <Joost.VandeVondele at mat dot ethz.ch> --- (In reply to Richard Earnshaw from comment #4) > No, it's not that simple. You're expecting us to try to debug a program we > know nothing about. We don't have time to do that. You claim this is wrong > code, but all you've shown is some output that is not what you're expecting. > How do you know this isn't wrong code in the program itself? Where's your > analysis of the code generated? As a now inactive gcc maintainer (and contributor of a few hundred bug reports), I am aware what the process is, and also that this is not the ideal test case. I mentioned we will try to work on reducing it, I hope that works. Maybe this filing the bug was bit premature, it is possible we have a coding error that so far only surfaced with with 15.2 on aarch64. Filing the bug was maybe a bit of an emotional response to seeing that our recent major release will be affected by our users on this combination. Our released binaries, which see tens of thousands of downloads for this architecture, are fortunately not affected. Even if this bug report is right now not helpful, we'll try to add to it as we get further with the analysis on our side.
