Repository navigation
feat(console): kmscon on the VTs, as on GNOME OS and Dakota - #1306
Conversation
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: hanthor The full list of commands accepted by this bot can be found here. Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
|
@hanthor this PR is back in your court ( Review asks:
Reply here or push a new head to this branch (same sign-off); CI and the review swarm re-run automatically and the PR returns to the normal path. Queued in your relay fix queue (8 waiting): fixes are handed out before new work as soon as you have capacity. |
Danathar
left a comment
There was a problem hiding this comment.
Stand-in review (the hive's reviewers are out of quota). Approving; don't merge until utah-packages#1157 (libtsm) and #1147 (kmscon) publish. The red contract is only check-repos on packages not in the stack's old pin 523507b8; rebase onto #1248/#1309. kmscon.conf, the 80-kmscon preset and the gdm drop-in are byte-identical to gnome-build-meta at the cited commit. Upstream's kmsconvt@.service has OnFailure=getty@%i.service, and getty@ is only disabled, so a kmscon that fails to start still ends at a login. Recovery and emergency mode use sulogin on /dev/console and are unaffected. The testsuite and e2e drive GDM and the initramfs, not a VT login. Worth the booted check you list: serial-getty@ttyS0 stays masked, so the VT is the only console login, and the fallback covers kmscon failing, not a black screen or hang.
There was a problem hiding this comment.
Summary: This change swaps the plain text login screens you reach with Ctrl+Alt+F2 and friends for kmscon, a nicer console that draws proper fonts, matching GNOME OS and Dakota. The setup itself looks right and matches upstream. But the kmscon package does not exist yet in the package source this image builds from, so the image cannot be built today and this should not land until that package is published.
intent-alignment
- Not ready to land: kmscon is not in the pinned factory yet (in-scope, medium). The linked issue's owner decision says "adopt kmscon once the factory builds it" (#1237), and the PR body says the same. At this head the
contractjob'sjust check-reposstep fails withNo match for argument: kmsconandNo match for argument: kmscon-pango(run 43812048338, job 133115662086, head0f2b93a). Those two names come from this PR atpackages/utah.toml:455-456andpackages/utah.toml:1027-1028. The factory import utah-packages#1147 merged without a kmscon recipe, and Fedora's spec hasBuildRequires: libtsm-devel >= 4.8.0, so a libtsm recipe must land first. Other misses in that same step (sudo-rs,oo7-*,kanata,dakota-config-*) come from the stacked base branch, not this PR.
Fix: no code change here. Land a libtsm recipe and then kmscon in utah-packages. Bump Utah's pinned factory digest. Rebase this PR on top of that and re-runjust check-repos.
What I checked and found correct: the contract paths match Fedora rawhide kmscon.spec 10.1.1 (%{_bindir}/kmscon, %{_unitdir}/kmsconvt@.service, %{_libdir}/kmscon/mod-pango.so in kmscon-pango). The package ships only kmscon.conf.example, so system_files/shared/etc/kmscon/kmscon.conf does not clash with it. Upstream kmsconvt@.service.in at v10.1.1 has Alias=autovt@.service, DefaultInstance=tty1, Conflicts=/OnFailure=getty@%i.service, so the claims in dakota-services.md and the preset comment hold. getty@ is only disabled, never masked (scripts/configure-services.sh:288, disable_unit at :28-30), so the OnFailure= fallback to getty still works.
Limit of this review: no image was built or booted. The tty1 handover between GDM and kmscon and the autovt@ alias override of systemd's own alias still need the booted check the PR body lists.
No findings from: correctness, security, style, docs-currency.
Confidence: 3/5 (needs attention) — 1 medium finding; 1 perspective requested changes
— hive: agent=reviewer backend=claude model=claude-opus-5-5 effort=medium claude=2.1.325
ee1afe8 to
b7a94cd
Compare
Install kmscon with its pango renderer and wire it as gnome-build-meta's gnomeos-deps/kmscon.bst does: kmsconvt@ enabled (DefaultInstance=tty1, autovt@ alias), getty@ and kmscon.service disabled, font-name=Adwaita Mono in /etc/kmscon/kmscon.conf, and gdm.service.d/kmscon.conf with Conflicts=/After=kmsconvt@tty1.service. Closes #1237 Assisted-by: Claude Opus via GitHub Copilot Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: James Reilly <jreilly1821@gmail.com>
There was a problem hiding this comment.
Summary: This change makes the text consoles you reach with Ctrl+Alt+F2 and friends use a modern console program (kmscon) instead of the old login prompt, matching GNOME OS and Dakota. The one blocker from the last review - the package was not yet available from Utah's package source - is fixed: the package source was updated and the availability check now passes. One code comment still describes the old setup and could be updated; it does not block.
Reviewed head 3feebff0. The compare from ec7f959 is 145 files, but nearly all of it is the rebase onto main; the PR's own diff is still the 11 files listed on the PR, and that is what I re-read.
Previous finding - fixed. [intent-alignment] kmscon/kmscon-pango not in the pinned factory: packages/utah-packages.repo:4 now pins sha256:d9972fab… (merged in #1372), and the contract job on this head (job 136142489407, head_sha 3feebff0) passed its "Check package availability in Utah's repositories" step (just check-repos). The factory spec (utah-packages packages/kmscon/kmscon.spec, 10.1.1) ships kmsconvt@.service and %{_libdir}/kmscon/mod-pango.so in kmscon-pango, which matches the contract paths.
docs-currency (low, advisory)
system_files/shared/usr/lib/systemd/system-preset/98-utah-enable-default.preset:50 still says "kmscon.service; its getty@.service line ships with kmscon". After this PR that is wrong: the factory kmscon ships no preset, and the getty@ disable now lives in 85-utah-desktop.preset. You already fixed the same wording in dakota-services.md. This line is not in the diff, so here is the patch:
-# 80-kmscon.preset (kmscon.service; its getty@.service line ships with kmscon)
+# 80-kmscon.preset (kmscon.service; its getty@.service line is in 85-utah-desktop.preset)Limits of this review: I could not check from source how the shipped kmsconvt@.service is installed (DefaultInstance=tty1, Alias=autovt@.service), so I could not confirm that enable * will spawn it on every VT. The PR body says the booted check (systemctl status kmsconvt@tty2, GDM on tty1) is still owed, and it should be done before release.
No findings from: correctness, security, intent-alignment, style.
Confidence: 5/5 (safe)
-- hive: agent=reviewer backend=claude model=claude-opus-5-5 effort=medium claude=2.1.327
|
Hive status — updated in place; edits do not notify. Full history is in the Activity Monitor. Review findings from This looks correct to me, with one stale comment left behind. What I checked against the tree at
Finding (low, docs-currency, unchanged line): Open question (not verified): the contract asserts |
kmscon draws the VT consoles, as on GNOME OS and Dakota. The factory publishes kmscon 10.1.1 (with libtsm) since pin
d9972fa(#1372), which is now on main. Rebuilt onto main's enable-by-default service model (#1233).What changed (gnome-build-meta
elements/gnomeos-deps/kmscon.bst@ 8a89846df11e)[parity]and[factory].parityaddkmsconandkmscon-pango. The pango renderer draws the configured font, and Adwaita Mono comes frombase.toml'sadwaita-fonts-all.files/kmscon/80-kmscon.preset.kmsconvt@.serviceis on through enable-by-default (DefaultInstance=tty1,Alias=autovt@.service,OnFailure=getty@%ias a fallback).getty@.serviceis disabled in85-utah-desktop.preset, because the factory's kmscon ships no preset and Hummingbird's90-systemd.presetenables getty@.98-utah-enable-default.presetalready disableskmscon.service. The contractenabled/disabledlists agree./etc/kmscon/kmscon.conf(font-name=Adwaita Mono) and/usr/lib/systemd/system/gdm.service.d/kmscon.conf(Conflicts=/After=kmsconvt@tty1.service) are byte-for-byte gbm'sfiles/kmscon/. The contract asserts them along with the binary, unit and pango module.serial-getty@ttyS0mask (desktop: serial getty spams journal on machines without a serial console #103) is unchanged.dakota-services.mdcovers the console; thekmscontriage entry now says it is installed from the factory; README/package-contract counts andsite/data/packages.jsonregenerated.Closes #1237
Verification
umask 022; just checkpassed on3feebff0(1413 tests).systemctl status kmsconvt@tty2after Ctrl+Alt+F2; GDM on tty1) is still owed.