IRC channel logs

<alganet>the kernels (builder-hex0-aarch64 and builder-hex0-riscv64) completed the stage0-posix bootstrap with minor changes. It was easier than I imagined.

<alganet>I will now spend some time investigating stage0-uefi, and possibly trying to hoan the bootstrapper to take the new builder-hex0 variants up to the mes build

<matrix_bridge><Andrius Štikonas> Uefi is tricker because you don't own entire memory...

<matrix_bridge><Andrius Štikonas> Alrhough I did get POSIX (amd64) version of mes to run on UEFI

<matrix_bridge><Andrius Štikonas> But to get more progress in emulating POSIX than i did, one need to have implement memory mapping

<matrix_bridge><Andrius Štikonas> And then parse Elf sections to place them in correct locations

<matrix_bridge><Andrius Štikonas> @irc_libera_alganet:stikonas.eu: https://git.stikonas.eu/andrius/stage0-uefi/src/branch/main/posix-runner/posix-runner.c can run mes on amd64

<matrix_bridge><Andrius Štikonas> But obviously this file is arch specific

<alganet>stikonas on stage0-uefi, I'm getting "fputs is not a defined symbol" just right after " +> amd64\artifact\cc_amd64.efi amd64\artifact\M2-0.c amd64\artifact\M2-0.M1"

<alganet>Am I doing something wrong there?

<alganet>1.9.1

<stikonas>hmm, let me try, haven't ran it for a long time

<stikonas>probably M2libc versions is wrong

<stikonas>alganet: hmm, ran to completion

<stikonas>alganet: next thing to check, did you run git submodule update --init --recursive

<alganet>gonna check the m2libc version, thanks! I'm doing it from the tarball, probably messed up something there, maybe it needs a pinned version

<stikonas>probably no submodule if it's git archive

<stikonas>I'm doing it from the repo

<stikonas>it's the same story with stage0-posix

<stikonas>it also needs submodules init'ed

<alganet>their tarball has the deps included pinned to the version https://github.com/oriansj/stage0-posix/releases/download/Release_1.9.1/stage0-posix-1.9.1.tar.gz

<alganet>but you're right! I will check with the git submodule flow, I think it's going to work

<stikonas>yeah, I know. I uploadaed that tarball

<alganet>it's a great help, submodules are fine but sometimes savannah git goes out and suddenly everything breaks

<stikonas>well, only mescc-tools are on savannah

<stikonas>and I think there is a copy on github

<stikonas> https://github.com/oriansj/mescc-tools

<alganet>probably a skill issue on my part, but I was never able to replace submodule origins in github ci

<alganet>I end up cloning and replacing the contents, that's why I ended up going for the tarball builds

<alganet>I eventually want to put stage0-uefi as it is on automated builds, like this one: https://github.com/alganet/abuild/actions/runs/23533620102

<stikonas>anyway, looks like mescc-tools is also pointing to github

<stikonas>anyway, I can probably generate tarball for 1.9.1 too

<alganet>I'll use the submodule path for now. The tarball would be great in the future, but it's not anything urgent

<alganet>confirmed, fputs was me using wrong files! Thanks for the help!

<stikonas>yeah, ok...

<stikonas>probably need to add some make target to get tarball

<stikonas>like stage0-posix has "make dist"

<stikonas>alganet: so it ran to completion?

<stikonas>the last thing it builds there is posix-runner.efi

<stikonas>though it has been 2 years since I've hacked on posix-runner...

<alganet>I'm probably stepping in my own toes here, but I'm not using its makefile directly. I'm trying to reuse the existing flow for abuild which generates the image using stage0. So now I'm bumping into short name MESCC-~1 FAT issues with my homegrown image builder

<alganet>that has nothing to do with stage0-uefi, it's entirely on my own

<alganet>host builds stage0 -> uses wrap to build image -> qemu boots completely stage0-built image

IRC channel logs

2026-03-25.log