[erikwb]

The internal speakers of the J713 (MacBook Air 13" M4): four SN012776 amps on the AOP's speaker stream, set up like the J616s, with the same protections.

Depends on #437 (jack), which depends on #434 (AOP and mics). The first nine commits belong to those two PRs; this one adds the last three. Rebased onto the reworked #434 (bootloader AOP handoff); these three commits are unchanged.

Commits

1. ASoC: dt-bindings: apple,macaudio: Add the J713 speakers
2. ASoC: apple: macaudio: Add the J713 speakers
   • The macaudio_j713_cfg uses SN012776 amps, one woofer and one tweeter per side, amp gain 15 and the −20 dB safe volume, with J616s-style AOP front-ends.
   • It adds a new speaker_fe_channels field (6 for the J713). macaudio limited the speaker front-end to one channel per amp, which conflicts with the AOP DAI's six channels and makes the open fail with EINVAL.
3. arm64: dts: apple: t8132-j713: Add the speakers
   • The four amps on i2c3 (from the ADT), each with its sense slots.
   • The shared SDZ line (AP GPIO 15) as the fail-closed SDZ group's fixed regulator, and the shared interrupt (AP GPIO 14).
   • The speaker card in the ADT provider order, with the amp-tx-zd-config idle masks.
   • The speaker and sense channels on the AOP audio node.
   • ps_i2c3 enabled.

No AOP driver change is needed: the J713 speaker stream is the J616s one.

From the ADT

	J713
Stream (iop-audio-controller/audio-speaker)	12 MHz, 250-bit frame, 48 kHz
Amps (/arm-io/i2c3/audio-speaker*, provider order = playback slot)	0x38 left woofer, 0x3b right woofer, 0x39 left tweeter, 0x3c right tweeter
Sense (speaker-config)	amp i: imon at RX byte 4i, vmon at 4i+2
Idle masks (amp-tx-zd-config rows 0/1)	0xf0f0 / 0x0f0f: the left woofer zeroes the right amps' sense, the right woofer the left amps'
SDZ / IRQ	AP GPIO 15 (function-reset) / AP GPIO 14, level low

The ADT's speaker entry lists a 4-slot playback mask and an 8-slot sense mask. The firmware's stream is nonetheless the J616s one: 6 playback and 12 sense channels, of which the four amps use the first 4 and 8.

I first wired it as 4/8, and the two mistakes cancelled out. Playback ran 1.5× fast and smeared each channel over both amps of its side (ch 0 and ch 2 → slots {0, 2, 4}). Reading 12-channel sense as 8 slowed it by the same 1.5×, so tones still showed up at the right frequencies. Timing a 3 s playback (2.1 s) gave it away. With 6/12, it runs in real time and every check below is clean.

Testing (J713)

All of this ran with speakersafetyd 3.0.0 holding the lease. The profile is the M3 Air's j613.conf with channels = 12: same chassis and speaker set. The ADT's per-unit DC resistances (4.00/4.10 Ω woofers, 3.55/3.68 Ω tweeters) match j613's nominals. The profile will go to speakersafetyd separately.

• Probe: all four amps bind to tas2764, and the SDZ regulator has 4 users (the fail-closed group).
• Interlock: without speakersafetyd, the volumes stay at the locked maximum (silent). With it, Speaker volumes unlocked. Stopping it gives Speaker volumes locked: Lock timeout.
• Silence + sense: 8 live sense channels at a few LSB around zero, frames whole, no stalls.
• Channel → amp → sense map:
  • Method: a −30 dBFS chord on channels 0–3 (500 / 700 / 1100 / 1300 Hz), with speakersafetyd's raw sense buffer dumped and each tone located by Goertzel.
  • Result: each tone appears in exactly one amp's I/V pair: ch 0 → left woofer, ch 1 → right woofer, ch 2 → left tweeter, ch 3 → right tweeter. Sense channels 8–11 are silent, with no cross-talk.
• Power: speakersafetyd's per-speaker power is plausible and distinct: 15.0 / 15.8 / 12.8 / 13.9 mW.
• By ear: left woofer low, right woofer low, left tweeter high, right tweeter high, as expected.
• No faults, no negative power, no DART or AOP errors.

Not done yet (userspace):

• A J713 asahi-audio graph. The tweeters have no passive crossover, so the desktop must not reach the raw speakers before one exists.
• UCM.
• The speakersafetyd profile.

On my machine the desktop is kept off the speaker card until those exist.

Checks:

• checkpatch --strict is clean on all three commits.
• dt_binding_check passes for apple,macaudio.yaml.
• dtbs_check on t8132-j713.dtb reports nothing for the amps, the card or the AOP.

[RyanTheTide]

The J713 setup preserves the four-amp shutdown group and volume lease, uses six playback slots and twelve sense channels, and keeps the first four playback slots in LW/RW/LT/RT order. The contributor's sense mapping and protected playback results are the J713 evidence.

This must not ship without speaker protection for J713, either an independently qualified speakerguardd #10 after #8 or a qualified J713 speakersafetyd profile. #10 is currently unqualified, and the speakersafetyd profile is still described as a separate submission. Pair the protection with UCM #6 and asahi-audio #10 so desktop playback goes through the crossover and the raw speaker path stays hidden. The tweeters have no passive crossover.

The stacked kernel at 5b6f5515e2f8 passed both full universal and Fedora-base WERROR builds (Image, modules and DTBs); both preserve all 131 common-config settings. ARM64 16 KiB KUnit passed 275/275. DCP host tests passed 63/63, and the PMGR and radio SID checks passed. The M4 host suite has the same five failing groups as the pinned base (18/23 on each); the AFKv2 HPD harness also fails on both. These are recorded failures, not a full host-suite pass. All six T8132 DTBs were checked: 13 schema diagnostics against 11 on the base, with exactly two new DCP dma-noncoherent binding errors from #429. The seven targeted binding checks passed. These software gates do not replace the required shared-code boots.

Merge after #434 and #437, with the shared audio boot checks completed. Keep the speaker shipping hold until that userspace/protection combination is ready.

[erikwb]

Superseded by #820, with the speaker commits unchanged (10–12). The speaker shipping hold you set is listed there, with the current state of speakerguardd#10 and #8, alsa-ucm-conf-asahi#6 and asahi-audio#10.