https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119951
--- Comment #3 from Chenyi Wang <invisparent at outlook dot com> --- (In reply to Andrew Pinski from comment #2) > The problem is the same as PR 109267. Can you try the patch in PR 109267 > which I just attached, it is against GCC 16 but this area in the file has > not changed since GCC 12 (r12-4616-g9262ae450d6a57). > > *** This bug has been marked as a duplicate of bug 109267 *** Hi Andrew, Briefly, your patch does work! Here are some details: When I switched to Linux(Ubuntu 24.04 LTS), use both gcc-13.3.0 and 14.2.0 (from apt-get install), I couldn't trigger this bug again, no matter what flags I've set, program always output 123. Knowing this bug may cause linkage problem (at least it's os-dependent), performing verification on Windows system is necessary. I chose gcc-14.2.0, one performs cross-compilation directly (original version), but another applied this patch first, then do cross-compilation (patched version). Finally I got compilers that --host,target=x86_64-w64-mingw32. Back to Windows, patched version behaves normally (outputs 123), with flags -std=c++17,23 -O0,1,2,3. But the original version only outputs 123 with -std=c++17 -O0 and -std=c++23 -O0,2,3, otherwise program crashes. So I do think this patch works. Extra info: binutils: v2.44 mingw-w64: v12.0.0
