Repository navigation
v0.4.0 — the z_ mark and Install LazyOS
v0.4.0 — the z_ mark and Install LazyOS
A minor release. The boot chain, the kernel and the compositor are the ones
v0.3.1 shipped; what changed is the project's face and the one menu
that writes a disk. The arcade button gives way to the z_ prompt, drawn in the
session's own Solarized and carried on every surface the system can draw; and the
root menu gains Install LazyOS, a front end to the installer this port already
has that reports the machine, hides itself when a write would be unsafe and reads
the result back instead of believing that the copy worked.
What changed since v0.3.1
- The mark is the
z_prompt. A cyanzand an orange cursor on a base02
tile, with the wordmarklazyOS_. The cursor orange is#e8743b, lifted from
#cb4b16so it passes 4.5:1 on base03 and can be used as text rather than only
as an accent.brand/README.mdis the brand itself - the mark, the wordmark,
the palette, the type and the rules - andbrand/lazyos.svgis the one source
every other file is rendered from. - The wallpaper is a keyboard in the dark with one lit key carrying the mark.
1366x912, the panel's own mode, so nothing is resampled; the wordmark,
lazy by designand the arcade continue line sit bottom left. The lock screen
keepsmountains.jpg: the desktop is the surface the brand owns, and the
photograph is what is up when nobody is looking. - Each surface asks for the form it can draw. The PNGs and the SVG where
pictures go (the launcher and About icon, the GTK theme's scalable copy, the
wallpaper), thez_in full and half blocks in fastfetch's logo and on the idle
screen (box and block characters lay out the same in every monospace font), and
plain text in/etc/issue. The idle screen centres line by line, so every row of
the art is padded to one width. scripts/make-brand.shrenders all of it withrsvg-convert, and ImageMagick
is no longer needed for the brand. The outputs are committed, because
build-rootfs.shstagesgit archive HEADand a build host without
rsvg-convertstill has to build; this release is the rerun of the script on
the build host thate707026left as a note, so the committed pictures are
librsvg's output of the same SVGs.- The project has a landing page:
site/plus.github/workflows/pages.yml,
published to GitHub Pages on every push that touchessite/,brand/or the
screenshots. JetBrains Mono is self-hosted (no third-party font request), and
the page carries its Open Graph and Twitter cards, JSON-LD,robots.txt,
sitemap.xml, a 404 page, the web manifest and the touch icon. The social
preview isbrand/social-preview.svg, 1280x640. - Install LazyOS is a front end, not a second installer (
rofi-install, the
root menu's red row andorg.lazyos.install.desktop). The target is fixed
(/dev/mmcblk0; a list of disks is a list of ways to erase the wrong one) and
the menu only reads before it offers anything: the room the copy needs (dfon
/and/boot, minus a build tree), the charger and the battery (below half
says so, below 20 % on battery refuses),/lib/modules/<release>+
vmlinuz-+initrd.img-(the kpart is built after the copy), the installer's
tools and asudothat would not prompt. Those rows carrynoneas their key,
so Enter re-checks instead of closing, which is what makes "Cannot install yet"
an honest row and not a dead one. Cancel is the first row and it is selected,
and the last gate is the disk's own name typed in full (mmcblk0). - The runner is the only writer, and what it writes with is the documented
installer.setsid lazyos-install-run TARGET REPOcalls
scripts/install-emmc.shonce withCONFIRM=yesafterenvand nothing else;
check-system.shFAILs ifparted,mkfs,cgpt,rsyncorddever appears
in that file, which is what makes "byte for byte the documented install" a test
instead of a promise. Around the call it does exactly the two things
disk_release()refuses to start with: stopudisks2and mask it for the
length of the run (stopping alone does not stop D-Bus from mounting p2/p3
mid-write) and unmount every mounted partition of the target. A failure in
either aborts before a byte is written, a trap restoresudisks2on every exit
path, andsystemd-inhibitholds a sleep lock meanwhile. - The copy is neither stoppable nor believed. A killed copy leaves a disk half
written, so no stop button is offered (like Format a disk and the apt install);
progress is the installer's own==>lines turning intobusycards under a
bar spinner, and the log can be opened live in a floatedfoot
(lazyos-install-log,less -R +Fwhile it runs, over
$XDG_STATE_HOME/lazyos/install-<timestamp>.log). Success is read back: the
threePARTLABELs one at a time andfutility showon p1 must say
Body verification succeeded; a failure carries the installer's last
ERROR:line, ANSI-stripped and un-prefixed, becausefail()puts one back.
Reopening the menu shows what the last run left (running,failed,
did-not-finish), and a stale pid is not taken for a running install. - The session gained the
lazyos-installwindow and the icon language that goes
with it:UI_I_SHIELDandUI_I_CHECKinlib-rofi.sh, the
system-software-installicon in the LazyOS theme, the drop-in that floats the
log, and05-system.md§ Install LazyOS in the manual.
Validation
The install path was exercised end to end on a loop device carrying the real pmOS
layout and a signed kpart copied from this machine's own kernel partition: the
success path (status done, the three labels and Body verification succeeded),
the failure path, the udisks2 window (inactive/masked-runtime while writing,
active/enabled after, no leftover mask), and the wizard's blocked, Enter-rechecks,
wrong-name, right-name and running-menu paths. That is how the empty "$@" that
made bash read DEVICE= as the command name - Failed to execute 'DEVICE=' - and
the doubled ERROR: prefix on the failure card were found. The seven install
checks of check-system.sh pass in the same harness.
The brand is a visual change and was checked as one: the four renders are the
output of scripts/make-brand.sh on this build host, and the two text forms come
out byte-identical. The card written from this build is the first real installation
test; docs/test-checklist.md § Install LazyOS lists what
to try, and scripts/check-system.sh runs on the booted card as always.
Assets
| asset | what it is |
|---|---|
google-asurada.img.xz |
6 GiB disk image: the three ChromeOS GPT partitions (kernel, /boot, root), ready for dd |
google-asurada-rootfs.tar.zst |
the same root filesystem as a tarball (with /boot inside), for writing your own card with scripts/write-sd.sh |
google-asurada.vmlinuz.kpart |
the signed FIT kernel partition of that image; it embeds the image's root=UUID, so it belongs to this image and no other |
SHA256SUMS |
SHA-256 of the three assets |
sha256sum -c SHA256SUMS
xz -dc google-asurada.img.xz | sudo dd of=/dev/mmcblk1 bs=4M conv=fsyncOn the Chromebook: developer mode, crossystem dev_boot_usb=1 dev_boot_signed_only=0, insert the
card and pick external disk in the firmware menu. It comes up to a greetd prompt — user
brunux, password debian, passwordless sudo. Change the password before using it for
anything real. scripts/install-emmc.sh installs it to internal storage, and on the card
Alt+Space → Install LazyOS is the same call with the questions asked first.
What it does not carry
The artifact is sanitized and asserted clean by scripts/lib-dist.sh:
no Wi-Fi profiles (a released image knows no network and asks for one), no SSH host keys and no
machine-id (both generated on first boot, so every installation has its own identity), and no
personal files beyond the session configuration this repository ships.
Known limitations
- Install LazyOS refuses to run from the disk it would overwrite, which is the
correct answer when the menu is opened on the eMMC install: it says so and
stops. The installation is made from the microSD, which can always be selected
again in the firmware menu. - The install cannot be stopped. Killing
rsyncormkfsleaves a disk that
looks written and is not, so the card explains that the copy is not interruptible
instead of offering a button that lies. - The image is 6 GiB (
IMAGE_SIZE_MIB=6144), which leaves about 2 GiB in the
root partition; a larger card keeps the space afterp3unused unless the
tarball is written withscripts/write-sd.sh, which sizesp3to the whole card. - The tarball's
/etc/fstabis the unconfigured one the build leaves behind:
write-sd.shgenerates the real one. - Artifacts are not byte-reproducible yet (FIT timestamp, filesystem UUIDs).
- The battery gauge's chemistry and time-to-empty properties are upstream failures;
waybar reads the capacity directly, which works. - There is no hardware video decoder, and Vulkan is not usable (PanVK answers Mali-G57 not
supported); the session renders through EGL and Panfrost, and video is CPU work.
Upstream
Vendored from pmaports commit 0e458a2417d6f5ed61c7b70621d5632e30136091 (2026-10-15), mainline
Linux 7.2.7, swayfx 0.6 on scenefx 0.5.