[fbalazsy]

J616s (MacBook Pro 16", M4 Pro, T6040) running Linux at EL2 under aura: hibernation and suspend-to-RAM with the desktop session kept. Builds on #1445, which is merged.

Base

• 258 commits on integration/aurora at aa68b3c. The first 168 are the series as it stood at d24d697, rebased. The base's move of the SoC words into per-SoC data (sleep.c) met three of the clean-up commits: the shared header now carries the base's definitions, the PMU write helper takes the USID as the base's does, and what the base reads from its match data (the PMU write command, the boot stage, the response vector's enable for the sleep) stays with the T6040 part as its own values. The 89 on top are new in this push and are listed under "New in this push".
• Every commit of mine carries its Signed-off-by; the two commits taken from others (RyanTheTide's phy: apple: t6040: tolerate a busy recovery domain, AxxzyWasTaken's spmi: apple: drop replies left by firmware at a firmware-driven resume) keep their authors and trailers.
• The aura side, J616s at EL2: hib=1 fixed layout, T6040 S2R windows, the 26.7 display handoff and the S2R wake path aura#148, is merged; aura main has it.
• Replaces J616s (M4 Pro): hibernation (suspend to disk) under aura's fixed boot layout #906.

What is in it

Area	Change
PM core	pm_hibernate_is_restored() lets restore callbacks tell a restored image from an aborted hibernation; a boot chain with a fixed layout (aura hib=1) admits hibernation.
arm64	Spin-table CPUs at EL2 are parked in a pen and released across S4 and S2R; EL2 state is restored for the hibernation resume.
soc/apple	aura hibernation layout; RTKit cold restart and re-adoption of a coprocessor found running or halted after a power loss; PMP (data SRAM kept across S2R), AOP, SEP, DockChannel.
soc/apple sleep	T6040 S2R as a part of the generic driver (sleep.c + sleep-t6040.c, struct apple_sleep_part in sleep-internal.h; one platform driver apple-sleep, the node enabled per board, on the J616s only). The entry takes the hardware down in a fixed order (CPU complexes, power gates, axi2af bridges, APCIe, DCS, the MasterMini S2R timer, then the commit word) and the wake brings the same blocks back, before syscore resume. The coprocessor and enclave registry stays in sleep-registry.c; the driver binds it once in probe and each SoC's part gives the two answers (T8140's rules and T6040's are the ones they had). APPLE_T6040_SLEEP stays an option, now the switch of the T6040 part under APPLE_SOC_SLEEP (with MFD_MACSMC=y). The part keeps its own suspend ops: suspend-to-idle passes through untouched and deep is offered only on a kernel booted with t6040_sleep.allow_entry=1. Four module parameters are left (allow_entry, dry_run, sleep_wd_secs, wd2_secs; 79 before); the progress marks, PMU records, probes, dumps and the experiments that were off by default are gone, the steps that were on by default are unconditional. Dry-run levels are 1 (every step that can be taken back, refused at the commit point) and 2 (also the last steps); the old level 1 never reached the commit point and is removed.
Display (drm/apple)	Internal and external T6041 display processors across S4 and S2R: the stop order the firmware expects, power words, DART streams, AFK endpoints paused and resumed, backlight level sent again. External links are given back before the sleep and set up again after it. The HDMI output's link is started once per plug, from the converter's power-up, as macOS does: a second start through an HPD pulse left the display black on some cold boots. A display that the wake's link has just woken can list the modes of its standby first: a Type-C display that lists other modes than before the sleep is read again once, and the hybrid pipeline is not handed over on the first lists after a wake. A display that fits one pipe again no longer counts as waiting for a second one.
GPU (drm/asahi)	G17 drain and restart around S4 and S2R, the GPU's PMGR word found through its PMP report entry.
USB / Type-C	A T6040 Type-C port keeps its xHCI host through a deep sleep (PHY stopped and started under it), a pin D dock port included while DP is down; USB4 routers get their Type-C PHY in the DT; ACIO, CIO PHY and tipd across sleep.
PCIe, NVMe, radio	DART locked tables re-imported; ANS/NVMe; brcmfmac survives a hibernation that was aborted or only thawed. After an S2R the internal host's two links are set up again in the noirq phase, with the port word, the tunables and the root-port link words the boot loader had set, so the PCI core finds the functions. The BCM4388's Bluetooth function comes back without its firmware: hci_bcm4377 sees that at resume (its configuration windows read reset values, boot stage 0), sends nothing and starts the transport's reset recovery at once instead of after command timeouts.
Camera (apple-isp)	T6040's ISP is powered down between streams and its firmware started again from its data segment as loaded (the T8132 scheme), so the camera works after a sleep.
ANE	H16 firmware 26.7.1, resident programs, system sleep and hibernation, on the base's hung-request recovery: its hung_stop and session quarantine are used unchanged, the hibernation events go through its notifier, and an engine quarantined at bring-up is powered off for a sleep (and left off at the resume) instead of refusing every sleep of that boot.
KVM	Resuming from disk is a cold boot, so the physical counter starts again while a VM's counter offsets stay: the guest's counter went backwards and its clock stood still. The offsets of every VM are moved by the lost count at syscore resume.
SMC	The sleep's commit writes WKTP = 2 as macOS does; after a full wake WKTP = 0 is written. Without it the SMC treats all the time awake as a power nap: it refuses manual fan control for the rest of the boot and asks for a sleep under load.
Other drivers	AIC3, pmdomain, SPMI, PMU RTC, pinctrl (every AP pin and the block's global word put back after S2R), ProRes T6041 across sleep (on #1260's variant: the J616s rows are its built-in ones, the per-Mac file stays an optional override), SIO restart, DockChannel HID, AOP ALS, cs42l84 jack interrupt.
DT	T6040 sleep controller, ProRes, JPEG, DP audio and USB4 PHY links on the J616s.
Clean-up (63 commits)	Comments, dead code and state nobody reads, per-sleep messages at debug level (the display driver prints 96 lines per wake instead of 141), names and named constants, pm_sleep_ptr() instead of #ifdef CONFIG_HIBERNATION in PM operations, dev_* messages from the T6040 part, the sleep node's unit address, bindings for apple,t6041-display-power and aurora,s2r-mailbox. One of them fixes a build configuration: with CONFIG_HIBERNATION=n a suspend-to-RAM stopped the display firmware and could not start it again; the restart is built for every system sleep now. For the comment, rename and constant commits the objects are identical before and after; sleep.o built without the T6040 part has every symbol unchanged from f259e90.

New in this push (89 commits)

Area	Change
SMC (4 commits at the end of the mfd group)	A suspend that is called off no longer leaves the machine without its fans. This SMC does not take MBSE = slpw back: after slpw and then waka with no sleep between, it stopped running the fans (0 rpm with the CPU die at 100 to 113 C) and refused a fan request until the reboot; after the whole commit (WKTP, slpa, the AP idle notice) and no power loss its sensor reports stood still as well, so the thermal zones read a constant. A dry run, a device that refuses, a wake event during the entry and a refusal at the commit point all ended there, on the previous head of this series too. On T6041 the whole sequence is now sent from the sleep controller's last stop, with interrupts off and every refusal behind it (apple_rtkit_ap_idle_polled() is new for that); on the other SoCs slpw moves from .prepare to the noirq suspend, right before the commit. A suspend that ends short of the power loss sends the SMC nothing.
USB3 after a suspend-to-idle (phy-apple-t6040)	The loss of USB3 until a reboot is fixed at its cause: in a suspend-to-idle genpd powered the PHY coprocessor's domains off under the running host, and the next stop of the coprocessor got AP idle not acknowledged: -62. The PHY now refuses that power-off while its firmware runs. With it: a powered-down sleep counts as the coprocessor's power cycle, a host whose USB3 is not back at a wake stays at USB2 and tries again at the next wake, and the CIO bring-up recovers the coprocessor and shuts the lanes down when it fails.
Display (drm/apple)	A resume commit no longer sends a mode set to an external pipe whose sink is reported gone (the mode set failed: -5 of every suspend-to-idle with an HDMI display); the hotplug that reports the sink back sets the mode. A T6041 sleep keeps to the pipe's own power words; a firmware that was given up stays stopped across an S2R, and a dead internal processor no longer refuses every sleep; the sleep registers are mapped once at probe.
Wake time	The two PCIe ports share the waits of a wake instead of taking 100 ms twice each (noirq resume 423 -> 118 ms); the dwc3 glue is asynchronous, so the USB hosts no longer start behind whatever the PM core's main thread waits for; the Type-C controllers do not wait 100 ms for a wake interrupt that has already come; a BCM4388 that lost its power is restarted from a freezable work item once the resume of every device is over (its resume callback: 619 ms -> 24 us).
Bluetooth	The core is resumed after the cold function's reset, not before it (sending frame failed (-19) at the first wake of a boot).
Hibernation	The jack codec, the RTC alarm, the pins, the AIC masks and the SPMI interrupt masks are put back after a resume from disk; tipd revalidates a port restored from disk itself, so the hibernation hook no longer has to write roles.
Thunderbolt / Type-C	No wait for the router stop in PM_POST_*, the LSTX level at a router stop keyed on T6040, the T6040 DP IN adapters kept disabled, no watch for a router whose node is disabled. Without a USB4 device here these ran as regressions only.
NVMe, ISP, ProRes, ANE	The RTKit object of a restore is configured as probe does; one predicate each for "this suspend takes the power"; timer_wakeups instead of irq_timeouts in the ANE's debugfs.
Tests	macsmc_sleep_order_test expects the lid read (it failed at the base); new KUnit cases for the SMC's last stop; the host tests of the PM callbacks follow the code they extract.

Hardware results

J616s (48 GB), firmware 26.7.1, aura main (9887a6b5) at EL2, kernel built from this tree with APPLE_SOC_SLEEP and APPLE_T6040_SLEEP enabled.

This head (278b7b9) is the rebase onto aa68b3c of the head the tables below were run on (4b0be95 on fdef45b), plus one commit (dmaengine: apple-sio: stop the T6041 firmware at the sleep's commit). On the rebase (3d13866) with that commit's module: cold boots, three suspends to RAM, one suspend to idle, a level 2 dry run and two sleeps refused at the enclave's request, with the panel and both external displays, Wi-Fi and Bluetooth back each time. Hibernation and the tables below were not run again.

On 4b0be95, one boot. Set-up: a Samsung G95NC in picture-by-picture with one half on the HDMI port and the other in DP alternate mode, a USB3 disk, charger; no dock. All three displays were seen lit after the boot and after every step.

Test on 4b0be95	Result
Cold boot	Panel and both external displays, Wi-Fi, Bluetooth, the USB3 disk; ANE at the first attempt; the SMC registered for the sleep controller's last stop.
Suspend-to-RAM (deep, RTC wake after 25 s)	3 of 3, one of them after the hibernation: the same compositor process, three outputs with their modes, Wi-Fi associated and answering, Bluetooth, the disk. "noirq resume" 118 ms, "resume of devices" 406 ms; no mode-set error. The SMC's sequence goes out at the last stop ("sleep warning at the last stop: 0 messages in 200 ms") and the wake record is the usual one.
Suspend-to-idle with the HDMI display	No mode set failed; the HDMI half is back with the firmware's hotplug, about 5 s after the wake.
Dry run level 2	Refused at the commit point; nothing is sent to the SMC, its reports move and it takes a fan request afterwards.
Hibernation from the running desktop, reboot into the resume	1 of 1: image restored, both external displays, the disk, Wi-Fi, Bluetooth, a fan request taken; ANE exact, AES equal in both directions, ProRes frames identical.

On the previous base (6970a51) the same commits ran over five kernels as they were written, most of it with a dock (hub, Ethernet, card reader), the USB3 disk and one USB-C DP display:

Test	Result
Suspend-to-RAM	34 in all, none failed; with the dock 12 USB devices on 9 SuperSpeed links after each.
The USB3 failure's recipe (cold boot, suspend-to-idle 15 s, then a dry run)	genpd's power-off of the PHY domains refused once per port, no -62, 9 SuperSpeed links before and after. Before the fix: 2 of 2 lost USB3.
PHY fault injection on the dock port (drop_next_power_down_ack)	Dropped acknowledge and a real S2R: 9 links. Dropped acknowledge and a dry run: the host stays, at USB2, nothing latched. A real S2R right after: USB3 back, 9 links.
Fans and SMC	Before the fix (the 81 commits without the SMC ones): a dry run left both fans at 0 rpm at 100-113 C under load and a fan request refused; in the same boot the SMC had run the fans before it. After it: reports live and a fan request taken after a dry run, after a refusal before the enclave's request (apple_sep sleep_refuse_before_send), and after an RTC alarm that fired during the entry.
Hibernation	Two in one boot, each restored, a suspend-to-RAM after them; an RTC alarm set before the first one fired at its time after the restore, one that passed while the machine was off was gone and a new one could be set.
USB disk across a suspend-to-RAM	Mounted read-only with a reader running: the same block device and SHA-256. Mounted read-write with a writer: 120 blocks of 1 MiB written with fsync across the sleep, all 120 read back with the right SHA-256, no btrfs or I/O error.
ANE, Bluetooth	A held ANE session exact across a sleep; Bluetooth powered off across a sleep and on again, two sleeps back to back: no error line.
Wake time, kernel only, from "Enabling non-boot CPUs" to "PM: suspend exit"	Nothing attached: 0.285 s (noirq resume 125 ms, devices 62-88 ms), the panel's firmware back at 0.37 s. HDMI + DP + disk: 0.76 s. With the dock: 2.7 s, of which 2.2 s is the dock's hub tree coming back level by level by reset-resume.

Not run on either of these two heads: anything with the dock (it was run on the previous base only), the KVM guest across the sleeps, a USB4/Thunderbolt device (none here). T8140 was not run by me on any head, and the SMC and wake-time commits change paths it shares: the noirq slpw on every SoC, device_enable_async_suspend() on the dwc3 glue, the BCM4388 restart, the skipped 100 ms in sn201202x_resume.

Review

The comments and commit messages of the 89 new commits follow the same rule as the rest since the last review: they say what the hardware or firmware does and what was measured on the J616s.

Checks

Run on 4b0be95 against its base fdef45b (not repeated after the rebase onto aa68b3c), on an aarch64 host with its own compilers (gcc 16, clang 23, rustc 1.99), not the pinned ones of the workflows.

Check	Result
tools/asahi/run-host-tests.py	324 runs, all ok with the host's compiler differences shimmed away (the base's new brcmf-recovery-tests among them). Without the shims 28 runs fail on this host's compilers; that run was not repeated at this base.
KUnit, C suites of the directories the series touches (qemu arm64)	1712 tests: 1703 pass, 4 skipped, and the five apple-display-gate-t6041-max cases that fail at the base as well (base: 1686 tests, those five and macsmc_sleep_order_test, which passes with the series).
KUnit, ANE boot.kunitconfig and transport.kunitconfig	230 of 230 each.
KUnit, drivers/soc/apple/sep.kunitconfig	605 tests, 603 passed, 2 skipped; in the workflow's form (LLVM, apple_sep*) 301 of 301.
scripts/dtc/apple-dt-check.sh (CHECK_DTBS on every Apple tree, dt_binding_check on the changed bindings)	No warning the base does not have (402 lines at both, 137 DTBs); the 12 changed bindings pass.
sep-source workflow: the tools/aurora-sep tests, the three DWC3/ATC compile profiles	The Python tests pass (14 lines, -O ones and the ABI step included); the three apple-dwc3-atc-compile.sh profiles compile; apple-sep.o builds.
checkpatch.pl --strict on the commits written for this push	Clean but for one "prefer kzalloc_obj" note in the brcmfmac commit.

Open

• A sleep refused at the commit point left the SIO firmware stopped until the next completed sleep: fixed by the last commit. The T6041 firmware now stops in the sleep's last step, as the T8140's does, so a refusal finds it running. After two refused sleeps and a dry run the AES engine behind it still encrypts and decrypts equal to the software implementation.
• The SMC's last stop cannot refuse. If the SMC does not answer there, each of its three writes waits its 5 s with interrupts off before the sleep goes on. An event the SMC sends during the 240 ms of the sequence is taken from its outbox and handled after the wake.
• After hot re-cabling (a full-width DP display on both pipes, the HDMI cable plugged while it ran, then the monitor switched to picture-by-picture) the HDMI half stayed dark once: "HDMI: no sink behind the converter yet, link started again (3/3)". A cold boot had both; the same switch of the monitor worked on the next two tries, one of them after a hibernation. Not reproduced, cause not known. A WARNING at typec_altmode_update_active() came once in the same re-cabling and was not looked at.
• A resume from disk still logs "Unable to change power state from unknown to D0, device inaccessible" for the internal host's functions; they work afterwards.
• The ANE's H16 firmware start with ignore_loglevel: unchanged from the previous description. Independent of sleep.
• A mounted USB disk is not carried through a hibernation (the hook that hibernates this machine takes the Type-C controllers down first).
• With a dock most of a wake is the USB core's reset-resume of the hub tree. usbcore.old_scheme_first=Y and the ports' fast-enumeration quirk took 0.64 s of it here with the same devices and device numbers; that is a setting, not part of this series.
• Questions the new commits leave for you: whether the T8140 SMC also stays in its dark-wake policy until WKTP is written 0, and whether it loses its fan policy after slpw + waka as this one does (then its noirq path has the same hole after the noirq phase that the last stop closes on T6041); whether sn201202x_resume may skip the 100 ms on every SoC with that controller; whether the AIC mask resync after a resume from disk should be registered for every SoC; whether a host left at USB2 by a wake should try a normal start at once instead of at the next wake. 263989e (DART in an off domain) is not in here.

How to try

Deep sleep stays opt-in, on the kernel command line:

t6040_sleep.allow_entry=1
echo deep > /sys/power/mem_sleep; systemctl suspend

With t6040_sleep.dry_run=1 (writable at run time) the same request stops at the commit point and resumes warm.

[RyanTheTide]

Before this can merge, please reword the comments that name macOS driver classes and methods, mark steps as RE or decompile-verified, or say where a value was read. That covers mainly t6040-sleep.c, plus t6040-usb2.c, dcp.c and macsmc.c. Each comment should say what the hardware or firmware does: the register, the bits, the order, and what you measured on the J616s. Values and code can stay as they are.

[fbalazsy]

Done in 7aefde0, thanks for the pointer to #1625.

• The comments that named host-driver classes or routines, marked a step as reverse-engineered, or said where a value was read (a trace, a recording, a run, a source file) now describe the hardware or firmware: register, bits, order, and what was measured on the J616s. Mainly t6040-sleep.c and dcp.c, also t6040-usb2.c, macsmc.c, nvme/host/apple.c, pmp.rs, t6040.dtsi and apple-ane.rst. The private labels of my bring-up notes went with them.
• Values and code are as they were: with comments stripped the files are identical, apart from six strings (five parameter descriptions and one log line) that carried the same labels.
• One real change came out of it: t6040-sleep.c still printed a list of steps that "had never run on hardware" at every suspend and needed i_know_untested to get past it, although all of them run. The list and the parameter are removed in the last commit; the commit is reached with no live blocker and t6040_sleep.allow_entry=1, and a requested dry run comes first.

Rebased onto aea2deb on the way. Tested on the J616s on this head (suspend-to-RAM, hibernation, suspend-to-RAM again; details in the description). That round was on battery without the dock, so the external displays have not run on this head yet; I will add that run.

What is left in t6040-sleep.c are the module parameters and records marked "Bring-up"; they go when the T6040 becomes match data in sleep.c.

[RyanTheTide]

Tested on a second J616s (M4 Pro, 24 GB, firmware of a macOS 27.0 install), Linux at EL2 under aura main 27298d21 (includes aura#148), kernel = integration + this series:

• Deep sleep, RTC wake after 30 s: 25 of 25 sleeps resumed, each a real commit and mailbox return (wake: resumes now +1 per cycle): 10/10 on c99d8b7, then 4 + 10 on 7aefde0 with the four commits below (two boots). After every wake: Wi-Fi reassociated, gateway ping, an 8 MiB NVMe write + fsync + read, internal panel and the USB-C DP monitor, 7 input devices, CPUs 0-13. Wall time 32.3-33.5 s for a 30 s alarm.
• One miss: on the first sleep of one boot, the dock port's USB3 side did not come back. phy-apple-t6040-usb2 392a90000.phy: RTKit: AP idle not acknowledged: -62 at the suspend, then PHY not back after the sleep (-5), dwc3-apple 392280000.usb: not resuming xHCI ... -5; the port came back USB2-only (hub on bus 3, the USB3 device gone) and stayed that way, with phymxwrap suspend failed: -11 on each later sleep. 24 of the 25 wakes had all 8 USB devices.
• One 603 s deep sleep on battery: about 0.34 W average from the fuel gauge (rough: the battery was full and the gauge corrects itself after a wake). Awake with the panels off: 3.9-4.1 W.
• Kernel log after each wake: the hci_bcm4377 -19 lines you list under Open, plus 1-5 apple_soc_edac 'event-log source 9' reports in the first ~10 s after the wake.

Four commits on top of 7aefde0 are on aurora-silicon/linux branch j6s2r/el2-sleep-fixups. Please pick them into this branch (or say if you want them changed):

1. soc: apple: t6040-sleep: leave suspend-to-idle alone: the PM core calls suspend_ops->begin/end for s2idle when there are no s2idle_ops, so t6040_sleep_begin() refused every s2idle with -EOPNOTSUPP once the controller probed. .begin now returns 0 for anything but PM_SUSPEND_MEM, .end only undoes a MEM .begin, and .valid offers MEM only with t6040_sleep.allow_entry=1, so mem_sleep keeps s2idle as the default.
2. soc: apple: share the sleep registry so one kernel builds both controllers: sleep.c and t6040-sleep.c both exported the nine apple_sleep_coproc_*/apple_sleep_enclave_* symbols, hence depends on !APPLE_SOC_SLEEP. The registry moves to sleep-registry.c; each controller binds to it in its probe. asahi.config gets CONFIG_APPLE_T6040_SLEEP=y, so one config builds the T8140 and T6040 controllers.
3. soc: apple: t6040-sleep: depend on a built-in SMC driver: built in, it calls apple_smc_read_u32/write_u32.
4. soc: apple: t6040-sleep: name the T8140 driver instead of its pull request: comment wording.

Checks on that branch: universal config builds (both controllers =y, and each one off), sleep KUnit 6 pass / 1 skip, checkpatch --strict clean, s2idle 3/3 with no kernel errors and the deep rows above on the J616s, s2idle 3/3 on an M1 Pro and an M2 with no new kernel errors.

[fbalazsy]

Picked up, thanks for the second machine and the four commits. Head is f259e90 on integration/aurora 107b361 (the brcmfmac .thaw commit dropped out, the base has it now). In order:

1. 7aefde0 as you tested it, then your four from j6s2r/el2-sleep-fixups, unchanged.
2. One fix for them: the registry commit moved apple_sleep_commit_stops_coprocs() and the enclave registration out of sleep.c, and tools/asahi/sep-suspend-errors-tests.py reads them from there, so it and its 21 mutation runs failed ("missing function"). The test now reads the three functions from sleep-registry.c.
3. The follow-up I had announced as a separate pull request, 35 commits; it was finished and it removes what you would otherwise have to review twice:
   • t6040-sleep's bring-up parameters, marks, records, probes and dumps are gone: 79 module parameters down to 4 (allow_entry, dry_run, sleep_wd_secs, wd2_secs); the same in the display, GPU, PCIe, pinctrl, PHY, dwc3 and PMP parts of the series.
   • dry_run=1 never reached the commit point (the PMU RTC refuses a deep sleep that is not firmware-driven in .prepare); it is removed and the levels are 1 and 2 now.
   • The T6040 controller is a part of the generic driver: sleep.c + sleep-t6040.c, one apple-sleep platform driver. Your sleep-registry.c and struct apple_sleep_ctl stay as they are; the driver binds the registry once in probe and the SoC's part supplies the two answers, each SoC with the rules it had. APPLE_T6040_SLEEP stays a real option (the switch of the T6040 part, with your MFD_MACSMC=y), so the line in asahi.config stays valid. Your suspend-to-idle change is kept as written: the part keeps its own suspend ops, .valid offers deep only with allow_entry=1.
   • Two comments that still cited pull request numbers are reworded.

If you would rather merge 9bee928 first and take the 35 commits as a second pull request, say so and I will split it there; they sit on top and nothing below them changed.

On the J616s here, kernel from this head with APPLE_SOC_SLEEP and APPLE_T6040_SLEEP (lid closed, no dock): cold boot, suspend-to-idle, dry run 1 and 2, suspend-to-RAM x3, hibernation, suspend-to-RAM after it; KVM guest, SIO AES and ProRes checked after each. The same list passed on 9bee928 earlier today. Host tests 292 -> 306 ok, KUnit with the same six failures as the base, Rust KUnit 573 of 575, sep-source, the three DWC3/ATC profiles and the dt check pass; details in the description.

Not verified: T8140 on the merged driver (its object differs from before only in the functions that now go through the part; a sleep on a T8140 would be welcome), and external displays, the dock and the kept USB hosts on this head. The USB3 miss on your dock port ("AP idle not acknowledged: -62") is under Open; I could not try for it without a dock.

[RyanTheTide]

@fbalazsy thanks for picking up the fixups and folding the T6040 controller into sleep.c. To merge this we need:

1. A rebase onto the current integration/aurora (028d762); the branch conflicts now.
2. Signed-off-by on 9bee928, c92a639, 9794539 and f259e90.
3. Your external-display run on the rebased head, and whether you have seen the dock port come back USB2-only after a wake (phy-apple-t6040-usb2 "AP idle not acknowledged: -62") on your machine.
4. Please reword "follows the order of macOS's quiesce and commit" in the description to describe the hardware order itself.
   One PR is fine with us; there is no need to split off the 35 follow-up commits. Once the rebased head is up we will run the J616s sleep rows and a regression pass on our other Macs.