Skip to content

Another m4cho trying to help - J616s - #58

Closed
fbalazsy wants to merge 1 commit into
aurora-silicon:mainfrom
fbalazsy:j616s/m4cho-real-sptm
Closed

fbalazsy wants to merge 1 commit into
aurora-silicon:mainfrom
fbalazsy:j616s/m4cho-real-sptm

Conversation

@fbalazsy

@fbalazsy fbalazsy commented Nov 2, 2026 •

Copy link
Copy Markdown
Contributor

Another macho on M4 trying to help: our J616s (MacBook Pro M4 Pro, T6040) path that keeps SPTM resident and runs Omarchy as a guest of a small EL2 monitor, so hardware TSO works for Linux processes. It complements the kcbz=0 boot object of #29.

On hardware today: Omarchy desktop under the resident SPTM, 14 CPUs, Wi-Fi, USB gadget, NVMe root, and tso_litmus across clusters (cpu1 E × cpu9 P, 40 M iterations) shows 15 reorderings with TSO off and 0 with the TSO bit on.

Not done: silent freezes after 60 to 330 s were traced to the monitor mapping through SPTM from guest state; the fix is in F37, which ran 9 min without a freeze; a rarer freeze is still being chased. Left USB-C xHCI sits on IRQ 0. DCP and external displays are not done. Details and the help-wanted list are in j616s-sptm/README.md.

No Apple code is in the tree. tools/kc_patch.py patches the user's own kernelcache (26.7.1 25G241, checked by SHA-256), and firmware and per-unit calibration come from the user's machine. See j616s-sptm/LEGAL.md.

The tree is the source of our hardware candidate F37 plus the F38 changes built since (KVM memslot delete fix, left USB-C ports held until their DART is bound). The kernel series carries aurora commits by Dj and Ryan Murray as its base, with their authorship.

An EL2 monitor that runs where XNU would, under Apple's resident SPTM, and
starts Omarchy as its guest at EL1, so hardware TSO can be enabled for Linux
processes. tools/kc_patch.py writes the monitor and the payload into the
user's own BootKC (26.7.1 25G241, checked by SHA-256); no Apple binary or
code is part of this tree. Reproduces candidate F37 byte for byte.

Work in progress: see README.md for the hardware status and the open
problems we need help with, and LEGAL.md for what the tree contains.

Signed-off-by: František Balázsy <frantisekbalazsy@gmail.com>
@fbalazsy
fbalazsy force-pushed the j616s/m4cho-real-sptm branch from 7f6023d to 5ce8691 Compare November 2, 2026 12:01
@fbalazsy

fbalazsy commented Nov 3, 2026

Copy link
Copy Markdown
Contributor Author

Closing: J616s work continues on the boot-object path in #66 instead of running under the resident SPTM.

@fbalazsy fbalazsy closed this Nov 3, 2026
dmitzsaz pushed a commit to dmitzsaz/aura that referenced this pull request Nov 6, 2026
…aurora-wip-20261028

Consolidate our m1n1 work on aurora-wip: J616s display handover, T8140 providers and j616s fixes
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant