IRC channel logs

2025-01-27.log

back to list of logs

<jleightcap>hey all, having fun poking through the bootstrapping sequence
<jleightcap>I'm sure maybe a reductive question, but what arches are "fully" supported?
<jleightcap>cc_x86 makes me wonder what the state of e.g. knight and rv32 are
<jleightcap>cc_x86 is incredible to read through btw, miraculous on its own
<Googulator>live-bootstrap only supports 32-bit x86 for now
<Googulator>Guix supports 32-bit and 64-bit x86 (with a cross-compiling step to go from 32- to 64-bit) officially, with riscv64 being worked on
<Googulator>(Guix has no kernel bootstrap)
<Googulator>Other archs are only supported within stage0, but not the larger bootstrapping projects building on them
<jleightcap>ah I see, it makes sense Guix is the reference target, thanks!
<jleightcap>my curiosity in asking is for poking at the below-firmware work
<jleightcap>i386, emulated or hardware, is out of my reach unfortunately
<oriansj>stikonas: we might have to add a version macro into M2-Planet's build so that mescc could do something like #ifdef __M2__ #if __version__ > 1.8 sort of thing
<oriansj>homo: nice to hear
<matrix_bridge><Andrius Štikonas> jleightcap: i386 code can run on x86_64 machines too
<oriansj>jleightcap: no architectures are "fully supported" (x86 tends to get the most testing because that is something what everyone has)
<matrix_bridge><Andrius Štikonas> riscv and amd64 just have some bugs in the middle layers
<oriansj>basically if you care about an architecture enough to report bugs, it'll probably keep working just fine. (If no one cares enough about an architecture to report regressions, it'll stop being supported)
<fosslinux>where is riscv broken?
<matrix_bridge><Andrius Štikonas> No reason we can't support bootstrap there, at least userspace
<fossy>amd64 is broken in mescc->tcc step
<matrix_bridge><Andrius Štikonas> @irc_libera_fosslinux:stikonas.eu: need tinycc build that has support for both riscv and x86
<fossy>ah, ofc
<matrix_bridge><Andrius Štikonas> If not that, it should work all the wau
<matrix_bridge><Andrius Štikonas> See https://gitlab.com/janneke/tinycc/-/commit/ea3900f6d5e71776c5cfabcabee317652e3a19ee
<matrix_bridge><Andrius Štikonas> It's just one commit
<matrix_bridge><Andrius Štikonas> amd64 us more broken
<matrix_bridge><Andrius Štikonas> (sorry for typos, writing from phone)
<fossy>i still understand you, all good ;)
<fossy>gotcha
<matrix_bridge><Andrius Štikonas> riscv could be progressed by conditionally applying that patch I guess
<matrix_bridge><Andrius Štikonas> Though it would be nicer to have unified codebase
<jleightcap>ah i386/x64 make total practical sense
<jleightcap>cross-compiling from 32 onto 64 is clever! does that come for free?
<janneke>snuik: later tell stikonas: oh that's great news!
<snuik>Will do.
<janneke>snuik: botsnack!
<snuik>:)
<malina>ye did that , as i think amd64 does'nt bootstrap yet or does it now?
<matrix_bridge><Andrius Štikonas> amd64 has bugs (probably in mescc)
<matrix_bridge><Andrius Štikonas> It can build tcc-mes binary on amd64
<matrix_bridge><Andrius Štikonas> And that binary can build some C files but crashes on some other files and hence can't rebuild 2nd tcc binary
<matrix_bridge><Andrius Štikonas> Nobody is actively looking at it right no as far as I know
<matrix_bridge><Andrius Štikonas> So feel free to investigate and debug
<janneke>stikonas, yeah guess that ekaitz (and efraim?) have been looking at that a bit while doing their riscv work
<janneke>but i believe this is still a know problem, it feels like "almost there"
<ekaitz>we have to make it better
<ekaitz>but i'm very busy :(
<matrix_bridge><Andrius Štikonas> Yeah, I do think that it is almost there too
<matrix_bridge><Andrius Štikonas> I'm quite busy too though
<matrix_bridge><Andrius Štikonas> Still behind with reviews for fossy...