/* evil! how this this happen */                                        \

239 if (slave_start_jmp_error_to_print != NULL) \ 240 gdm_error (slave_start_jmp_error_to_print); \ 241 gdm_error ("Bad (very very VERY bad!) things happening in signal"); \ 242 _exit (DISPLAY_REMANAGE); \ 243 }

I'd hate to be the poor soul who encounters this error message and finds no information about it except the source code.

It will only occur if everything is so irrevocably on fire that you need to debug a lot of code anyway, at which point the error message is useless.

Specifically, the branch is only reachable if the PID is not what was expected, or the location that the macro was supposed to longjmp to is not set, which would be fatal defects.

Yes, that may be it. Great find! And sorry to Sun Microsystems, it wasn't you this time.