https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=285401
--- Comment #63 from Mark Millard <[email protected]> --- (In reply to Dimitry Andric from comment #62) On main [so: 15 as stands], even though the link does not fail or even complain the result is still not operational: # ./environ-_progname ld-elf.so.1: /lib/libc.so.7: Undefined symbol "environ" # readelf -drs environ-_progname | grep -eenviron -e__progname 24: 0000000000230798 8 OBJECT LOCAL DEFAULT 24 __progname 25: 00000000002307c8 8 OBJECT LOCAL DEFAULT 26 environ 49: 0000000000000000 0 FILE LOCAL DEFAULT ABS environ-_progname.c # file environ-_progname environ-_progname: ELF 64-bit LSB executable, ARM aarch64, version 1 (FreeBSD), dynamically linked, interpreter /libexec/ld-elf.so.1, for FreeBSD 15.0 (1500034), FreeBSD-style, with debug_info, not stripped In other words: ports may build but fail to operate when building for main. In that sense, what happens on stable/14 and (from what I understand) on releng/13.5 makes it easier to know about a problem existing. My guess here is that llvm19's linker should be reporting like happens for stable/14 and releng/13.5 rather than prodcing a broken result. May be some part of its checking the status needs to have some operating-system specific code checking extra, OS specific details. Is there a way to have a test 'bulk -a' run based on stable/14 or on releng/13.5 that scans for the build failures --or at least for unusual build failures for further review? -- You are receiving this mail because: You are the assignee for the bug.
