Repository navigation
Description
Drafted with an AI assistant (Claude Code) from this machine's logs; I reviewed it before posting.
Mac and installed build
Model / board / SoC: MacBook Air 13" M3 (2024) / apple,j613 / apple,t8122 (driver profile T8122/J613)
uname -r: 7.2.11-14-1-ARCH
linux-aurora 7.2.11.aurora14-1, m1n1-aurora 1.6.1.aurora22-1, omarchy-aurora-firmware 20261014-7
omarchy-mac 0.2.3-2, omarchy-mac-boot 20261104-1
libfprint-git 1:1.94.100.r10.g6f9479c-5, fprintd 1.94.5-2
Distribution / local patches: Omarchy on Arch Linux ARM, no local patches
Module options: xart_writes=1 provision_keybag=1 (omarchy-mac-touchid.conf), xart_guard default (level 1)
xART lane: /chosen/apfs-preboot-uuid, no os_uuid_* on the command line
Both firmware versions and provisioning
asahi,os-fw-version: 14.7
asahi,system-fw-version: 27.0
Own-Mac provisioning status: calibration carved from this Mac's own iBoot System Container by omarchy-mac; identity keybag created by this install on its first Touch ID boot (2026-11-03)
Reproduction, expected result, and actual result
- 2026-11-03: turn on Touch ID in Omarchy, which creates the keybag on Linux's first boot with the driver. Enrol one finger. Verify works for the rest of that boot.
- Boot macOS once.
- Boot Linux again.
Expected: the Linux enrolment keeps matching after a macOS boot, or the driver recovers without lowering the guard.
Actual: on every Linux boot since the macOS boot (4 boots so far, 2026-11-05 to 2026-11-06), the xART guard refuses the enclave's rewrite of a kind-4 record and DESIGNATE_KEYBAG gets no reply. Matching and enrolment then stay blocked for the whole boot: diag/matching reads blocked and touchid reads not-ready. Verify is refused before the sensor is touched, so re-enrolling doesn't help either. Suspend and resume make no difference.
On the first boot, the guard's exception allowed the enclave to commit the existing kind-4 locker "during this install's first keybag creation". Every later boot still reports 0 created by Linux, so the same record counts as foreign once that exception no longer applies. The xART max revision went from 3226 (working boot) to 3350 (first broken boot) across the macOS boot. It is 3353 now.
No Linux packages changed between the last working boot and the first broken one.
diag now: abi=2 protocol=variant5 boot=warm xart=enabled xart_guard=1 xart_guard_refusals=1 keybag=present keystore=open restore=unknown sensor=failed sensor_attach=not-ready touchid=not-ready matching=blocked
Relevant error lines
Working boot (2026-11-03):
apple_sep: xART guard level 1: 10 session records exist, 0 created by Linux; the other 10 are protected
apple_sep 29e400000.sep: xART: APFS-resolved .gl extent (base 0x4d4000); 170 slots, 12 live records, max revision 3226, 0 malformed, 0 duplicate, 0 repaired; writes ENABLED
apple_sep 29e400000.sep: xART: using /chosen/apfs-preboot-uuid
apple_sep: xART guard: allowed the enclave to commit an existing kind-4 locker during this install's first keybag creation
apple_sep 29e400000.sep: sks: identity keybag provisioned
apple_sep 29e400000.sep: sks: LOAD_KEYBAG reply mailbox status 0, response size 84, copied 84 bytes
apple_sep 29e400000.sep: sks: snapshot identity bound to recovered and designated bag; generated lookup false
apple_sep 29e400000.sep: scrd: cold-match credential established true
apple_sep 29e400000.sep: sbio: pre-restore special unlock true, SCRD match context true
apple_sep 29e400000.sep: sbio: cold-match preparation true, component restore false
apple_sep 29e400000.sep: enrol: opt-in opt-in completed user catacomb state Some(7)
apple_sep 29e400000.sep: enrol: opt-in opt-in post-user master catacomb state Some(7)
Every boot after the macOS boot (current boot shown):
[ 10.024014] apple_sep: xART guard level 1: 10 session records exist, 0 created by Linux; the other 10 are protected
[ 10.024063] apple_sep 29e400000.sep: xART: APFS-resolved .gl extent (base 0x4d4000); 170 slots, 12 live records, max revision 3353, 0 malformed, 0 duplicate, 0 repaired; writes ENABLED
[ 10.850449] apple_sep 29e400000.sep: xART: using /chosen/apfs-preboot-uuid
[ 13.205047] apple_sep 29e400000.sep: sks: LOAD_KEYBAG reply mailbox status 0, response size 84, copied 84 bytes
[ 13.857020] apple_sep: xART guard: refused to rewrite a kind-4 record this Linux install did not create
[ 15.266059] apple_sep 29e400000.sep: sks: DESIGNATE_KEYBAG got no reply after 2060 ms
[ 15.266652] apple_sep 29e400000.sep: sks: DESIGNATE_KEYBAG did not complete; enrolment will refuse
[ 17.313762] apple_sep 29e400000.sep: sbio: pre-restore special unlock false, SCRD match context false
[ 22.498027] apple_sep 29e400000.sep: sbio: cold-match preparation false, component restore false
[ 27.617024] apple_sep 29e400000.sep: sbio: identity list unavailable after Catacomb restore with sensor registered
[ 27.617346] apple_sep 29e400000.sep: matching unavailable this boot: cold-match preparation, SCRD credential, Catacomb restore, or identity proof did not complete; template state is unproven
[ 32.738026] apple_sep 29e400000.sep: sbio: the sensor did not wake at attach
[ 38.370136] apple_sep 29e400000.sep: sbio: attach-time sensor bring-up did not complete; matching unavailable; retrying in 10 s
... sensor: the enclave did not register the sensor (5 late retries)
... sensor: authenticated attach incomplete after the late retry; capture unavailable until an enrol or verify brings the sensor up
[ 428.651010] apple_sep 29e400000.sep: verify: refusing before touching the sensor; cold-match prep, restore, or device-view identity proof did not complete. This is not a non-matching finger
Userspace:
touchid-activate: Touch ID is not ready to enrol and match: a match could not save its state (matching blocked)
fprintd: Device reported an error during verify: verification failed: the secure enclave refused a step (driver status 2)
Related issues, pull requests, or code examples
- apple_sep: on a J416c (M2 Max) the xART guard refuses identity-keybag creation, and the enclave stops answering #209 (J416c): the same guard refusal of a kind-4 locker. Its update says one macOS boot cleared the refusal there. Here a macOS boot after a working Linux setup is what started it.
- apple_sep: J316s enrolment from 11.36 fails on 7.2 under the default xART guard #269 (J316s): the guard refuses the post-match rewrite of an enrolment it counts as foreign, and
xart_guard=0fixes it. - M3 Air (J613/J615) Touch ID: SEP and Mesa values from the ADT, and a test machine #79: an earlier J613 report of bring-up wedging after the guard refuses a record a previous install left.
Workarounds tried: rebooting Linux several times, and suspend/resume. Neither recovers. I have not tried xart_guard=0, because of the macOS data-loss risk (FileVault is a possibility on this Mac).
Questions:
- Should the kind-4 locker this install adopted during its first keybag creation count as Linux-created from then on?
- Is there a supported way to recover that doesn't need
xart_guard=0? For example, deleting the keybag and turning Touch ID off and on again, or another macOS boot.
What was not tested
xart_guard=0(deliberately avoided).- Removing the keybag and enrolling again.
- Booting macOS a second time to see whether it clears the refusal, as in apple_sep: on a J416c (M2 Max) the xART guard refuses identity-keybag creation, and the enclave stops answering #209.
xarm_trace=1/sensor_trace=1. I can run these on request.- J615 or any other Mac.
Activity
techgyani commentedon Nov 8, 2026
Another data point, from a 14-inch M3 Max (
apple,j514m,apple,t6034) with the same Omarchy Touch ID path (omarchy-mac 0.2.3-2,libfprint-git 1:1.94.100.r10.g6f9479c-5,fprintd 1.94.5-2), guard at the profile default (level 1). Three macOS boots after a working Linux setup did not reproduce this issue here.Unreleased build: kernel
8eddf5aeb6ac+ local J514M patches (aT6034/J514MSEP profile copied fromT8122with the J514M sensor, and J514M device-tree nodes) + a local guard change described below; m1n1c6fef490+ omacom/m1n1#29.asahi,os-fw-version: 14.7,asahi,system-fw-version: unknown. omarchy-mac's eligibility check does not list this board, sotouchid-enableran with a local copy oftouchid-eligibilitythat addst6034 && j514m.The guard change: once per boot, at the first
DESIGNATE_KEYBAGof a keybag that already existed, and only when the gigalocker's max revision at probe is newer than the last revision Linux wrote (kept in a host file), the guard may let the enclave rewrite one existing kind-4 record that the enclave read this boot. Never a delete, never adopted into Linux's list, logged. It was added for this issue; on this Mac it was never used (below).touchid-enable)sks: identity keybag provisioned; norefusedline; the #212 first-setup window was not usedenroll-completedverify-match/verify-no-matchTouch ID: restored 1,verify-match,xart_guard_refusals0closed unused, norefusedline;Touch ID: restored 1,verify-match,xart_guard_refusals0bioutil -cunchanged (3), Touch ID unlock and a Touch ID prompt, login keychain unlocked; FileVault onsudothrough PAMEach Linux boot after macOS logged:
(The other two times: 53481 vs 53451 and 53517 vs 53487.)
Record headers from
apfs-xart-inspect.pybefore and after each macOS boot: macOS rewrote the same three kind-4 records each time (and, in the third boot, one kind-3 record), plus both roots. Lengths unchanged, no records added or removed. Linux's own boots changed only the roots.So on this Mac, the unlock after a macOS boot did not need a rewrite, unlike the J613 here. These logs don't show what differs.
The guard change, against
8eddf5aeb6ac(sep.rs,sks.rs,xart_guard.rs,xart_store.rs; applies withpatch -p1). It was written from the driver source at that commit and the logs in this issue and on this Mac; it contains no Apple code. KUnit under QEMU arm64: 131apple_septests passed, 8 of them new. On hardware, only the closed-unused path above has run; the rewrite path has not.Patch (guard window)