Re: kern.version and uname -v
- In reply to: Warner Losh : "Re: kern.version and uname -v"
- Go to: [ bottom of page ] [ top of archives ] [ this month ]
Date: Tue, 09 Jan 2024 20:23:36 UTC
I personally feel that building releases reproducibly is quite important. How can we trust the current 14.0-RELEASE images? Would be nice to be able to build them locally and know they are the same. What do we need to do to make sure that 14.1-RELEASE is built reproducibly? Thanks! PS: It looks like freebsd-reproducibility@ may have some issues? Jan 9, 2024, 03:29 by imp@bsdimp.com: > > > On Mon, Jan 8, 2024 at 7:17 PM Eugene Grosbein <> eugen@grosbein.net> > wrote: > >> 09.01.2024 4:16, Warner Losh wrote: >> >> > On Mon, Jan 8, 2024 at 1:58 PM Eugene Grosbein <>> eugen@grosbein.net>> <mailto:>> eugen@grosbein.net>> >> wrote: >> > For ages, "uname -v" output (sligtly polished sysctl kern.version) had the following format: >> > $ uname -v >> > FreeBSD 13.2-STABLE 36a037f15 KERNELIDENT >> > >> > This is the reproducible format: only include data that is identical from build to build. It's relatively recent (FreeBSD 11 maybe) >> > >> > Where KERNELIDENT is GENERIC for x86 distribution media. >> > >> > But now: >> > >> > # uname -v >> > FreeBSD 14.0-RELEASE #0 releng/14.0-n265380-f9716eee8ab4: Fri Nov 10 05:51:26 UTC 2023 root@releng1.nyi.freebsd.org:/usr/obj/usr/src/i386.i386/sys/GENERIC >> > >> > This is the old, non-reproducible format. We've had both formats for several major releases, and this format, with various tweaks as we went from CVS -> svn -> git. >> > >> > Maybe the problem here is that in the run up to 14.0 we didn't turn on reproducible builds? >> >> You meant opposite, did you? >> > > I don't think so. The quoted value is the old WITHOUT_REPRODUCIBLE_BUILDS format. We forgot to turn on WITH_REPRODUCIBLE_BUILDS in the branch before the release. > > Warner >