https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119376
--- Comment #32 from Andrew Pinski <pinskia at gcc dot gnu.org> --- (In reply to lucier from comment #31) > I'm trying to analyze some failures I'm seeing where > > (a) GCC 14 does not "optimize" tail calls with -foptimize-sibling-calls with > a system that expects this optimization, and it seems we eventually hit some > limit. > > (b) An earlier version of GCC 15, where our code attaches the musttail > attribute to appropriate procedure calls, gave error messages like > > ../include/gambit.h:11038:2: error: cannot tail-call: other reasons > > confirming that those tail calls Can you file that seperately. And we will try to analyze what is going wrong. Most of the time "other reasons" happens is due to the musttail call not being analyzed due to other code after it and we don't give a good answer for that. The other reason is due to having setjmp in the function and we don't produce a good diagnostic for that (I had a fix for diagnostic for GCC 16 but was waiting until 15 was branched off; though it might need some rework due to recent tail call fixes).
