On Mon, 21 Oct 2024 23:07:27 +0000 (UTC) RVP <r...@sdf.org> wrote: > On Mon, 21 Oct 2024, Sad Clouds wrote: > > > Yes, once I applied this patch > > http://cvsweb.netbsd.org/bsdweb.cgi/pkgsrc/shells/ksh93/patches/patch-src_lib_libast_features_syscall?rev=1.4 > > > > The test succeeds, but gcc returns error "redefinition of 'getcwd'" > > which is more of a NetBSD + SSP issue. > > > > OK, refreshing my memory again, I see the purpose of that ksh93 patch: it's > to make ksh use the syscall (which on NetBSD is SYS___getcwd--for some reason) > instead of rolling its own. So, that patch alone w/o Christos's SSP patches > won't work if you compile with -D_FORTIFY_SOURCE=2 and any -On flags.
Yes that is correct. When I looked at the sources, it had a bunch of ifdef statement, depending on the test results of getcwd. I believe the gcc error will occur either with or without the ksh patch. I simply ran the failing gcc command by hand with "-O2 -D_FORTIFY_SOURCE=0" and got past the failure. Other people hitting similar issues on NetBSD-9.4 may not have the technical skills for such workarounds. So I do think it needs fixing by NetBSD developers. Thanks.
