Repository navigation
m1n1-aurora: 1.6.1.aurora22-1, the J516C display and GPU defaults - #1707
Conversation
Pin omacom/m1n1 286529075cb5e56b574e9c8f9ebbe407e0b8de9e (omacom/m1n1#19, on omacom#18's merge): on the 16-inch M3 Max with the 40-core GPU (J516C), m1n1 requests its existing GPU handoff by default, subject to the existing checks. Nothing else changes in the source or the recipe. The qualification record stays build-only: hardware_tested is false. A native J516C GPU also needs the kernel's J516C board data, T6031 module loading and the native display enables, outside this package. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
[Medium impact] The packaging changes appear safe to merge as an edge candidate, not as a qualified rc/stable release. SummaryUpdates
No new actionable issues were established. Reviews (3) · Last reviewed commit: "m1n1-aurora: pin the merged J516C defaul..." · Reviewed by Greptile |
Pin omacom/m1n1 5d17a0348745e755299f2c6455f6f1c31236d66b, the reviewed head of omacom/m1n1#19. On the 16-inch M3 Max with the 40-core GPU (J516C), m1n1 now requests its existing display handoff by default as well as its GPU handoff, each subject to the existing checks. An owner's chosen.asahi,t6031-dcp=0 or chosen.asahi,t6031-gpu=0 keeps that one off; an existing gpu=0 still gets the display default. The version stays 1.6.1.aurora22-1: that version was never built or published from this branch. The qualification record stays build-only: hardware_tested is false. Native J516C graphics also need the kernel's automatic J516C display request and display-before-GPU gate (omacom/linux-aurora#246) and the kernel's J516C board data, outside this package. An installer image needs its boot.bin rebuilt from this package before its first boot carries the defaults. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
maralcbr
left a comment
There was a problem hiding this comment.
The diff is fine: version bump, _commit, the qualification JSON hash and the README.
Two things before merge:
- Please repin
_committo the m1n1#19 merge commit onintegration/auroraonce it's merged, not the PR head5d17a034(same as we did for #18). - Merging this publishes to edge, and
update-m1n1rewrites boot.bin on every J516C there. Please land it together with the linux-aurora pin carrying #246 + #249, so edge never pairs the new stage 2 with the old kernel. The J516C cold boot doesn't test that mix.
Holding until the J516C cold boot. I'll add build-approved after that.
|
Agreed. The body now makes both prerequisites explicit: repin to m1n1#19's actual |
scottjones
left a comment
There was a problem hiding this comment.
Thanks Chris. This is the same position as our review of omacom/m1n1#19: we're happy to approve after one J516C cold boot with omacom/linux-aurora#246, omacom/linux-aurora#249 and this m1n1, installed the way the release installs it (empty m1n1.conf, the release's 14.8.3 OS firmware stub). Thanks as well for the M3 Pro, M1 and M1 Air regression boots you posted on omacom/linux-aurora#246.
- No change on our 13" M3 Air (J613). We built this recipe repinned locally (m1n1-aurora 1.6.1.aurora21.test5d17a03, from 5d17a034) and installed it with a linux-aurora test kernel carrying omacom/linux-aurora#246, omacom/linux-aurora#249 and some other open PRs. The omarchy-mac-boot M3 Air leaf said "nothing to write", there was no /etc/m1n1.conf, and stage 2 reported 5d17a034. In /chosen, t8122-dcp, -gpu, -gpu-power-standin and -gpu-fuse-leakage were all 1 and t8122-dcpext was absent. The GPU came up as Apple M3 (G15G C0) on asahi, eDP ran at 2560x1664 on apple-drm, and there were no DART or DCP faults.
- The pin is 5d17a0348745, a commit on the omacom/m1n1#19 branch. If it gets squash-merged or its branch is deleted, the pin needs moving to the merged commit.
- This has to ship together with omacom/linux-aurora#246 and omacom/linux-aurora#249 and a matching installer rebuild, as you already said. On its own it's harmless, since the J516C just stays on the boot framebuffer.
|
@maralcbr The owner-reported J516C defaults-only cold boot is now posted for the exact supplied packages, including the normal greeter/desktop on packaged Mesa. I have linked the report and its one-machine/local-Stage1 limits in this description. Please review it for the cold-boot prerequisite. The remaining delivery order stays the same: m1n1#19 merge, repin this recipe to that actual integration merge commit, then coordinate with the kernel package carrying #246 and #249 before publication/image rebuild. No pins, approval labels or merges were changed in this follow-up. |
|
The J516C cold boot covers the prerequisite. m1n1#19 is merged as @chriskearney please repin |
Pin integration merge bc79acc4 after m1n1#19 landed, including the separately reviewed and boot-tested T6030 dependency fix in omacom#17. Update the qualification checksum and describe the combined pin without claiming a new native build or hardware run. Keep the package build-only and retain the normal maintainer approval and coordinated kernel delivery requirements. Generated-by: OpenAI Codex
|
@maralcbr the repin is pushed at The matching kernel pin is now #1735, |
maralcbr
left a comment
There was a problem hiding this comment.
Approving at 4b915212. The recipe and the qualification JSON both pin bc79acc4 (the m1n1#19 merge), and the JSON hash checks out. Next to the cold-booted 5d17a034, the only addition is m1n1#17: a T6030 display-DART request now brings its firmware-memory handoff with it. The memory-only and neither-flag paths are unchanged, a refused memory handoff can't publish the display marker, and T6031/T8122/M1/M2 never reach this path. Eryk's J514S external displays and your J516S boot cover it.
For the M3 Pro release notes: with dispext=1, dcpext=0 on its own no longer opts out. Set both to 0 to skip the external-display handoff.
Delivery order: kernel and Mesa go first in #1736, which carries the same two pins as #1735 and #1737 in one PR and is already building. Once #1736 is published and read back on edge, I'll add build-approved here and merge, so edge never pairs this stage 2 with the old kernel. After that, the exact J516C and M3 Pro boots on edge.
Pins
m1n1-aurora 1.6.1.aurora22-1tobc79acc434e612ffcb95bfc7cb2fd7543afac5f2, the actual integration merge of omacom/m1n1#19. On J516C (16-inch M3 Max, 40-core GPU), Stage2 requests the existing display and GPU handoffs by default, subject to the existing admission checks. The merge also includes m1n1#17: T6030 external-display handoff requests its required memory handoff.The recipe source pin, qualification source pin and qualification checksum agree. The README describes the merged source and its validation limits. Build flags, artwork, dependencies and installation behavior are unchanged. The version remains
1.6.1.aurora22-1; the earlier PR-head package was a private test build, not a repository release.Validation
git diff --checkpass at the new pin.hardware_tested:falseandstage1_auto_enrollment:falseremain unchanged.82f7d7c921f1d2a5a26583484a8d9292dc321a51, independently matched to the clean merge of chore: sync AUR packages #17 integration and chore: sync AUR packages #19. Relative to the previously tested chore: sync AUR packages #19 source, only chore: sync AUR packages #17's T6030 dependency fix and its host regression are added.602b92e6…, Stage25d17a034-dirty. It passed guarded combined-kernel regression boots on M3 Pro J516S, M1 J293 and M1 Air J313, followed by exact restoration: results.ca6c77a0+ Add snd-hda-macbookpro-dkms for pre-T2 MacBook and iMac speakers #24979f78916. Empty m1n1.conf, no enable override/modules-load workaround, standalone cold power-on, both chosen defaults, native display, GPU autoload/READY and packaged-Mesa SDDM/Hyprland passed. This is one existing local Stage1 and one reference machine, not stock-Stage1 or fresh-installer qualification.bc79acc4package has not yet been natively built or booted. Prior runs do not qualify this exact package or the full M1/M2/M3/Neo promotion matrix.Opt-outs, recovery and upgrades
chosen.asahi,t6031-dcp=0andchosen.asahi,t6031-gpu=0independently suppress their respective handoffs. To suppress both, set both.chosen.asahi,m3-defaults=0adds no defaults and preserves owner lines, including explicit=1. This package does not rewrite/etc/m1n1.conf, enroll Stage1, or change other board admission.Recovery instructions for a pre-Linux GPU identity-read hang explain the Recovery-side boot.bin opt-out, Stage1 precedence and all-graphics-off fallback. Keep a known-working original;
boot.bin.oldis not automatically a known-good backup. Persist the selected switches in/etc/m1n1.confafter Linux is available.Existing
install-aurora-sep.shMax installs may freezeupdate-m1n1; its--m3-handoffvariant explicitly setsgpu=0. This update does not migrate those installs. With kernel #246, existingdcp=1,gpu=0installs can gain the native display while the GPU remains off. The next Stage2 applies only afterupdate-m1n1rebuilds boot.bin and that image is booted.Delivery
Keep publication coordinated with a kernel package pin containing linux-aurora #246 and #249, including the carried #253 changes if selected by that kernel pin. Marcelo requested exact-candidate J516C and M3 Pro boot coverage before or immediately after edge delivery; the older owner cold boot did not include #253. No new hardware run is claimed here.
Maintainer build approval, package merge/publication and installer rebuild remain separate. No build-approval label is added by this repin. Do not release the loader paired with the old kernel. The installer must rebuild its own boot.bin from the matching package cohort for its first boot to carry these defaults; a package pin does not update an existing image. RC/stable promotion still requires the full qualification record and matched Stage1/Stage2, repair and upgrade checks.
Paired package review: kernel #1735, Mesa #1737 and this loader PR form the new trifecta pin. Keep their native builds, exact-cohort boot checks and delivery coordinated.