Date: Thu, 15 Feb 2024 21:37:37 +0100 From: Jo Durchholz <jo@durchholz.org> To: virtualization@freebsd.org Subject: Won't contribute KVM build :-( (was: Contributing the build of a KVM image) Message-ID: <b0a99787-8b7d-4931-9b9c-317158651ea8@durchholz.org> In-Reply-To: <dad74102-853b-4738-abd2-30b98b73ab33@durchholz.org> References: <dad74102-853b-4738-abd2-30b98b73ab33@durchholz.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On 29.01.24 23:41, Jo Durchholz wrote: > Hi all, > > where do I turn to if I want to contribute to the FreeBSD build/release > process so that a KVM image will be built? > > This is not actually about virtualization per se but about the > build&release process, so I guess I'd be wrong here. I'm sorry to report that I can't do it, due to lack of feedback and incomplete/outdated documentation. Current roadblock is that I don't see a list of valid value for TARGET and TARGET_ARCH. After much searching, I found that "make targets" will give me that. Trying "make.py targets" -> complains about unset MAKEOBJDIRPREFIX. Some Internet searches later I know it's just a temporary build directory, so I put it in ~/obj, export MAKEOBJDIRPREFIX=~/obj -> complains: Could not infer value for $XCC. Either set $XCC or pass --cross-bindir=/cross/compiler/dir/bin "which clang" gives me /usr/bin/clang, so why does it not find clang? Now trying src/tools/build/make.py --cross-bindir=/usr/bin targets Could not infer value for $XCPP: /usr/bin/clang-cpp Ehm... sure, no, there is no such thing as a clang-cpp. I have a clang-cpp-14. Not sure why there's no clang-cpp, https://clang.llvm.org/docs/UsersManual.html says it should be there, but then again the Debian package maintainer may have run into naming conflicts and decided to give a version-specific name. I could obviously use XCPP=/usr/bin/clang-cpp, but given that I haven't gotten an answer to my previous question and that I have to expect more roadbumps ahead, plus I'll have to learn the freebsd Arcanist tooling just to submit a patch... well, it's been death by a thousand papercuts, and sorry, that's more work than I'm willing to invest, and it's a KO anyway since I'm making something highly security-critical and I'm not confident I can learn enough of FreeBSD to make no stupid mistakes. So... Linux it will be. It's sad, because FreeBSD has a pretty impressive security track record that I'd really like to leverage, but if I can't achieve the traction to bring that advantage to the road, it's not useful for me - I'd need a competent FreeBSD person for that, and I don't have one, and my own competence seems insufficient. Please not that THIS IS NOT A CRITICISM OF FREEBSD. It's just a single data point, for anybody who wants to know what kinds of things hold back FreeBSD adoption; if there's no need for that (possibly because other things have higher priority, or a gazillion of other possible reason), please pretend I never wrote this. And, thanks for making FreeBSD. I know it's a lot of work. Regards, Jo
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?b0a99787-8b7d-4931-9b9c-317158651ea8>