[erikwb]

The T8132 AOP (J713, MacBook Air 13" M4). It takes the T6040 configuration, and the bootloader hands it over the T6040 way. This supersedes the AOP half of #434. The J713 audio that runs on it is in #820.

• dt-bindings: apple,t8132-aop on apple,t6041-aop.yaml. On T8132 the bootloader also programs the AOP DART's DAPF and reserves its windows. The firmware reaches those windows through its DART at their physical addresses, so they come in memory-region along with the firmware segments: up to 18 regions on T8132. T6041 stays at 4.
• soc: apple: aop: a match entry, HW_CFG_T8132 = HW_CFG_T6040.

The #434 review

• Loader ownership. Following the review, there's no ADT reader, DAPF write or raw iommu_map in the kernel, so the probe error paths that left DAPF writes and mappings behind are gone with them. The IOMMU core maps the reserved regions, and check_fw_regions() verifies them as on T6040.
• The handoff:
  • J713: hand the AOP over with its DAPF, firmware segments and windows m1n1#47 programs the 16 DAPF slices, reads them back and restores them on a mismatch. It reserves __ETEXT, the moving __OS_LOG and the 15 merged windows with identity iommu-addresses, and disables the AOP on any failure.
  • The window-union fix from that review is in 1b9b611ba3ef, with both counterexamples as regression tests.
  • arm: apple: reference m1n1's J713 AOP reservations from the OS tree u-boot#13 carries the reservations into per-kernel DTBs.
  • Both are merged on those repos' aurora-wip, but not yet on main. With an older m1n1 the AOP gets no memory-region, and the driver refuses to start it.
• Control DTB: m1n1's bundled tree takes the AOP node from the J713 DTB in the audio PR. That path, and the per-kernel DTB through u-boot#13, both worked in Omarchy's packaged layout (m1n1#47).

Testing

On a J713, with the audio PR on top, on integration/aurora 41c4055. The boot chain has the m1n1#47 and u-boot#13 handoff.

• Startup: the AOP's memory-region lists 17 regions (__ETEXT, __OS_LOG and the 15 windows), and RTKit: Initializing (protocol version 12). No DART faults, and the kernel stays untainted. The microphones, jack and speakers all run on it, and a full snd-soc-aop reload brings them back.

• Start failure and rebind:

  • Unbinding apple_aop times out waiting for the endpoints to acknowledge their shutdown (5 s, mask 0xfff).
  • On the rebind, RTKit endpoint discovery fails (−62).
  • The failure path then halts the AOP again, keeps its buffers until reboot, and fails the probe cleanly: no oops, untainted.

  So on J713 the AOP core can't be rebound without a reboot, but its failure path is safe. T6040 takes the same path. I haven't seen how a J616s behaves there.

Checks:

• checkpatch --strict is clean on both commits.
• These two commits build on their own, drivers/soc/apple with W=1. aop.o also builds in the universal WERROR build (defconfig + asahi.config).
• dt_binding_check is clean for apple,t6041-aop.yaml.

[erykwieliczko]

Rebased: commit message only (not pushed yet)

#819 still merges cleanly. But since 28174f88b ("soc: apple: aop: Write the T6040 firmware's own nCal boot argument"), the T6040 configuration writes nCal 1, and HW_CFG_T8132 = HW_CFG_T6040 inherits it. The message of soc: apple: aop: Add T8132 still said nCal 0.

I reworded it to nCal 1, with a pointer to 28174f8, and added my sign-off with an [eryk: …] note. The code is unchanged, and Erik stays the author.

Testing:

• On a J713, with nCal 1, on 21f7ffc with this, ASoC: apple: J713 (M4 MacBook Air) microphones, headphone jack and speakers #820 and other PRs: the AOP gets its 17 regions and starts, and the microphones, speakers and jack card come up on it. Details are in ASoC: apple: J713 (M4 MacBook Air) microphones, headphone jack and speakers #820.
• It builds alone on 21f7ffc with W=1.

I haven't pushed it. If you're fine with it, I'd force-push it to j713/aop-t8132 together with the #820 rebase.

---

🤖 Written by Claude Opus 5.5 (in Claude Code) for Eryk; the rebase, fixes and tests above were done the same way.

[erykwieliczko]

Merged. This comment is from Claude, working for Eryk, not from Eryk himself.

Rebased onto integration/aurora 565c8025e715 and force-pushed first, as Erik agreed. No code changes: only the commit message of soc: apple: aop: Add T8132 changed. It now says nCal 1, which the T8132 config inherits from the T6040 one since 28174f88b, with an [eryk: …] note.

---

🤖 Written by Claude Opus 5.5 (in Claude Code) for Eryk.