Re: git: a098111a28ed - main - secure: Add ssh-sk-client to all consumers of libssh
- In reply to: Shawn Webb : "Re: git: a098111a28ed - main - secure: Add ssh-sk-client to all consumers of libssh"
- Go to: [ bottom of page ] [ top of archives ] [ this month ]
Date: Fri, 25 Apr 2025 00:56:06 UTC
On Thu, Apr 24, 2025 at 6:00 PM Shawn Webb <shawn.webb@hardenedbsd.org> wrote: > > On Thu, Apr 24, 2025 at 11:56:03PM +0100, Jessica Clarke wrote: > > On 24 Apr 2025, at 23:45, Shawn Webb <shawn.webb@hardenedbsd.org> wrote: > > > On Tue, Apr 22, 2025 at 02:07:50AM +0000, John Baldwin wrote: > > >> The branch main has been updated by jhb: > > >> > > >> URL: https://cgit.FreeBSD.org/src/commit/?id=a098111a28ed59e1ab1101ad09913f0235ebd28f > > >> > > >> commit a098111a28ed59e1ab1101ad09913f0235ebd28f > > >> Author: John Baldwin <jhb@FreeBSD.org> > > >> AuthorDate: 2025-04-22 02:05:28 +0000 > > >> Commit: John Baldwin <jhb@FreeBSD.org> > > >> CommitDate: 2025-04-22 02:05:28 +0000 > > >> > > >> secure: Add ssh-sk-client to all consumers of libssh > > >> > > >> These all failed to link with ld.bfd used by GCC due to > > >> Fssh_sshsk_sign being an unresolved symbol. > > >> > > >> Fixes: 65d8491719bb ("secure: Adapt Makefile to ssh-sk-client everywhere") > > > > > > Hey John, > > > > > > I'm getting the following error from the RTLD when the rc scripts > > > start sshd: > > > > > > ==== BEGIN LOG ==== > > > ld-elf.so.1: /usr/lib/libprivatessh.so.5: Undefined symbol "Fssh_sshsk_sign > > > ==== END LOG ==== > > > > > > This is on HardenedBSD 15-CURRENT/amd64. I'll try to reproduce next > > > week with vanilla FreeBSD, unless someone else beats me to it. > > > > I don’t understand how this is meant to work. sshsk_sign is used by > > sshkey.c, which is in libssh, so why are we putting the definition of > > sshsk_sign (namespaced) in each and every program? The original motivation was to mimic: https://github.com/openssh/openssh-portable/commit/7b47b40b170db4d6f41da0479575f6d99dd7228a In order to sync our Makefiles with upstream Makefile.in, as yet another binary is about to come in OpenSSH 10.0 (sshd-auth). Having the same order, even the same line breaks facilitate adapting to changes. I don't mind reverting it if other downstream projects/linkers are having issues. I'm holding on to another related change (D49801) because of a098111a28ed. We can always try again after OpenSSH 10.0 has been merged (or not). > I'm not sure. Looking more closely at this commit and the commit > referenced on the Fixes: line, I'm wondering if the use of $() is > valid instead of ${}. It should expand (maybe a style.Makefile(5) infringement, again to match as close as possible Makefile.in). > > Thanks, > > -- > Shawn Webb > Cofounder / Security Engineer > HardenedBSD > > Signal Username: shawn_webb.74 > Tor-ified Signal: +1 303-901-1600 / shawn_webb_opsec.50 > https://git.hardenedbsd.org/hardenedbsd/pubkeys/-/raw/master/Shawn_Webb/03A4CBEBB82EA5A67D9F3853FF2E67A277F8E1FA.pub.asc