Skip to content
Closed
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
360 commits
Select commit Hold shift + click to select a range
52e740f
mfd: macsmc: drop a duplicate include and a never-NULL drvdata test
fbalazsy Nov 12, 2026
2f8d83f
mfd: macsmc: report restore failures with dev_err()
fbalazsy Nov 12, 2026
613cb63
mfd: macsmc: build the hibernation callbacks with pm_sleep_ptr()
fbalazsy Nov 12, 2026
2313aeb
power: reset: macsmc-reboot: log power-off and reboot at usual levels
fbalazsy Nov 12, 2026
70b3c01
irqchip/apple-aic: drop two dead conditions around the sleep syscore
fbalazsy Nov 12, 2026
40b3695
pinctrl: apple: describe the resume order by what the pins need
fbalazsy Nov 12, 2026
9719ac0
spmi: apple: say why the masks are rewritten in the early restore
fbalazsy Nov 12, 2026
e57c498
pmdomain: apple: pmgr-pwrstate: set restore_noirq with pm_sleep_ptr()
fbalazsy Nov 12, 2026
aa404cd
dmaengine: apple-sio: report restart failures with dev_err()
fbalazsy Nov 12, 2026
c17e76f
pinctrl: apple: report the per-sleep pin restore at debug level
fbalazsy Nov 12, 2026
652cfe2
spmi: apple: report the early mask rewrite at debug level
fbalazsy Nov 12, 2026
c9afea3
pinctrl: apple: drop the restored-pin report and never-true tests
fbalazsy Nov 12, 2026
4368700
spmi: apple: set the restore callbacks with pm_sleep_ptr()
fbalazsy Nov 12, 2026
6c5a94a
soc: apple: Kconfig: tidy the T6040 sleep and aura hibernation texts
fbalazsy Nov 12, 2026
adb570c
soc: apple: sep: hold no sleep for a restore the notifier refuses
fbalazsy Nov 12, 2026
3678af9
rtc: apple-pmu: have the alarm set again after a resume from disk
fbalazsy Nov 12, 2026
9eabdf5
pinctrl: apple: put back every pin the driver set after a resume from…
fbalazsy Nov 12, 2026
993255e
spmi: apple: drop replies left by firmware at a firmware-driven resume
AxxzyWasTaken Nov 12, 2026
1d474d8
spmi: apple: rewrite the masks once after a resume from disk
fbalazsy Nov 12, 2026
ecea7d3
mfd: macsmc: write WKTP only to an SMC known to take it
fbalazsy Nov 12, 2026
ec3e925
pinctrl: apple: write a pin back in steps only on T6041
fbalazsy Nov 12, 2026
0105206
mfd: macsmc: make the WKTP values constants
fbalazsy Nov 12, 2026
34c02e1
irqchip/apple-aic: mark the T6041 AIC as lost in a firmware suspend
fbalazsy Nov 12, 2026
03837bb
irqchip/apple-aic: resync the masks only after a resume from disk
fbalazsy Nov 12, 2026
3f9a32b
irqchip/apple-aic: resync the masks after a resume from disk on any SoC
fbalazsy Nov 12, 2026
66a2c2d
mfd: macsmc: share the SMC handshake between probe and restore
fbalazsy Nov 12, 2026
6dfa5af
spmi: apple: share the interrupt reset between probe and restore
fbalazsy Nov 12, 2026
58fe936
mfd: macsmc: expect the lid read in the sleep order test
fbalazsy Nov 12, 2026
316b800
mfd: macsmc: leave the SMC's sleep commit out of a dry run
fbalazsy Nov 12, 2026
f487ada
mfd: macsmc: say when a sleep is refused after the SMC's commit
fbalazsy Nov 12, 2026
1995d17
mfd: macsmc: warn the SMC of a sleep only where it is committed
fbalazsy Nov 13, 2026
56cf925
mfd: macsmc: send the sleep commit from the sleep controller's last stop
fbalazsy Nov 13, 2026
70e4cd3
dmaengine: apple-sio: stop the T6041 firmware at the sleep's commit
fbalazsy Nov 13, 2026
3f8f167
mfd: macsmc: leave out the rest of the last stop for an SMC that does…
fbalazsy Nov 13, 2026
51eafe8
soc: apple: sleep: order the first stops explicitly, drop an unused e…
fbalazsy Nov 13, 2026
ae28dfc
soc: apple: sep: probe again through the aura reprobe helper
fbalazsy Nov 13, 2026
db76911
soc: apple: sleep-t6040: decide the refusal before the setup writes
fbalazsy Nov 13, 2026
9e87510
soc: apple: sleep-t6040: require the register blocks at probe
fbalazsy Nov 13, 2026
38ceef9
soc: apple: sleep-t6040: map each power word page once
fbalazsy Nov 13, 2026
7e81a56
soc: apple: sleep: keep the registry's walkers with the registry
fbalazsy Nov 13, 2026
5066af2
soc: apple: sleep: keep the registry's answers in the registry
fbalazsy Nov 13, 2026
f750948
soc: apple: sleep: keep the wake information with the registry
fbalazsy Nov 13, 2026
c60ff39
soc: apple: sleep: build every sleep controller and drop two symbols
fbalazsy Nov 13, 2026
9af13df
soc: apple: sleep: leave the SoC with a part of its own out of the la…
fbalazsy Nov 13, 2026
2ff613c
soc: apple: sleep-t6040: look the register blocks up by the shared names
fbalazsy Nov 13, 2026
11aa7cc
soc: apple: sleep-t6040: share the test for the GP host's system domain
fbalazsy Nov 13, 2026
8c022da
mfd: macsmc: tell drivers that an image is being loaded
fbalazsy Nov 13, 2026
56f3993
soc: apple: let the platform refuse hibernation through a hook
fbalazsy Nov 13, 2026
beeff17
soc: apple: sep: learn a cold resume from the PM core, not from the PMP
fbalazsy Nov 13, 2026
cd45353
soc: apple: drop the restore callbacks' checks for an abandoned hiber…
fbalazsy Nov 13, 2026
db9d742
mfd: macsmc: own the pmVC request for the VDD_CIO rail
fbalazsy Nov 13, 2026
7fbe928
pmdomain: apple: pmc: put the voter enables back after a resume from …
fbalazsy Nov 14, 2026
852b403
soc: apple: PMP, AOP, SEP, store and DockChannel across hibernation a…
fbalazsy Nov 10, 2026
329b95c
HID: dockchannel, iio: aop_als: J616s keyboard, trackpad and light se…
fbalazsy Nov 10, 2026
3589dbd
soc: apple: pmp: keep the T6041 PMP's data SRAM across S2R
fbalazsy Nov 10, 2026
6ecabf6
ASoC: cs42l84: keep the jack interrupt quiet across system sleep
fbalazsy Nov 10, 2026
9670dfd
soc: apple: pmp: keep the T6040 PMP bound as the T8140 one
fbalazsy Nov 10, 2026
d9e9867
soc: apple: pmp: drop the sampling of the firmware's text across a sleep
fbalazsy Nov 12, 2026
2baa22a
soc: apple: aop: describe the cold restart without test notes
fbalazsy Nov 12, 2026
228804e
soc: apple: pmp: describe the restore and the T6041 SRAM without logs
fbalazsy Nov 12, 2026
7f2b22a
iio: aop_als: describe the sleep handling by what the sensor needs
fbalazsy Nov 12, 2026
c88809a
soc: apple: dockchannel: set the restore callback with pm_sleep_ptr()
fbalazsy Nov 12, 2026
bfb8238
HID: dockchannel: build the hibernation callbacks with pm_sleep_ptr()
fbalazsy Nov 12, 2026
e491222
ASoC: cs42l84: set the codec up again after a resume from disk
fbalazsy Nov 12, 2026
bcff068
soc: apple: dockchannel: filter short receive IRQs only after a restore
fbalazsy Nov 12, 2026
b25063a
soc: apple: dockchannel: skip set_wake again
fbalazsy Nov 12, 2026
681bd79
soc: apple: probe again through the aura reprobe helper
fbalazsy Nov 13, 2026
ee74b39
soc: apple: learn a cold resume from the PM core, not from the PMP
fbalazsy Nov 13, 2026
3f5a6bf
soc: apple, ASoC: cs42l84: drop the restore callbacks' checks for an …
fbalazsy Nov 13, 2026
282f733
iommu, nvme, PCI, brcmfmac: Apple T6040 storage and radio after a pow…
fbalazsy Nov 10, 2026
3b5ba29
PCI: apple: set a conventional host's ports up again after a real S2R
fbalazsy Nov 10, 2026
3f16a93
nvme: apple: describe the T6040 sleep as it runs now
fbalazsy Nov 10, 2026
564e430
Bluetooth: hci_bcm4377: restart the transport of a chip a sleep left …
fbalazsy Nov 12, 2026
d069087
PCI: apple: bring a conventional host's links up in the noirq phase a…
fbalazsy Nov 12, 2026
59788bd
PCI: apple: set a conventional host up after a real S2R without switches
fbalazsy Nov 12, 2026
f66b5c3
nvme-apple: one prefix rule for the step functions
fbalazsy Nov 12, 2026
1236dc1
iommu/apple-dart: drop a quoted log from the restore comment
fbalazsy Nov 12, 2026
99c8a1d
iommu/apple-dart: report the locked stream re-import at debug level
fbalazsy Nov 12, 2026
d7b3b36
iommu/apple-dart: set the restore callback with pm_sleep_ptr()
fbalazsy Nov 12, 2026
cf6ed82
Bluetooth: hci_bcm4377: resume the core after a cold function's reset
fbalazsy Nov 12, 2026
75f367a
nvme-apple: configure the RTKit object of a restore as probe does
fbalazsy Nov 12, 2026
9e51c28
nvme-apple: one flag for a controller the sleep controller stops
fbalazsy Nov 12, 2026
1d123f0
nvme-apple: say what the T6040 ANS does around a sleep
fbalazsy Nov 12, 2026
5bccc0d
Bluetooth: hci_bcm4377: a table for the boards restarted in place
fbalazsy Nov 12, 2026
19b11dd
Bluetooth: hci_bcm4377: keep the core's PM notifier off a cold function
fbalazsy Nov 12, 2026
0c9fefe
Bluetooth: hci_bcm4377: report a function the sleep reset at debug level
fbalazsy Nov 12, 2026
15ef9b4
iommu/apple-dart: decide at probe which DART gets its tunables again
fbalazsy Nov 12, 2026
1fad760
PCI: apple: let the ports share the waits of a wake from S2R
fbalazsy Nov 13, 2026
dc0b939
PCI: apple: wait after the training only for a link above 5 GT/s
fbalazsy Nov 13, 2026
b634b7b
wifi: brcmfmac: pcie: restart a function back from a power loss behin…
fbalazsy Nov 13, 2026
80cfcb3
PCI: apple: drop the unreachable T6040 MSI slice branch
fbalazsy Nov 13, 2026
f1eaa5b
PCI: apple: fail the probe on a malformed port tunable
fbalazsy Nov 13, 2026
774f371
PCI, nvme: apple: word the comments as hardware contracts
fbalazsy Nov 13, 2026
e8660e1
PCI: apple: name the restore helpers after what they do
fbalazsy Nov 13, 2026
5459bca
PCI: apple: select the conventional restore by match data
fbalazsy Nov 13, 2026
16fa403
PCI: apple: check the RID-to-SID table after a restore
fbalazsy Nov 13, 2026
e66bd72
iommu/apple-dart, nvme-apple, PCI: apple: drop the restore callbacks'…
fbalazsy Nov 13, 2026
e408512
PCI: apple: say why the sleep reads the root port words each time
fbalazsy Nov 13, 2026
6f80d31
PCI: apple: bring the conventional ports up side by side after a resu…
fbalazsy Nov 13, 2026
f0617f9
wifi: brcmfmac: pcie: restart behind the restore after a resume from …
fbalazsy Nov 13, 2026
151cfbb
wifi: brcmfmac: do not sleep 500 ms per interface in the suspend
fbalazsy Nov 13, 2026
4dcdbe8
thunderbolt, phy: apple, usb: typec: J616s USB4 and Type-C across sleep
fbalazsy Nov 10, 2026
e88c115
phy: apple: t6040: write the lane tunables again after each DP link p…
fbalazsy Nov 11, 2026
3a6a1db
phy: apple: t6040: shut the PHY down when four-lane DP powers off
fbalazsy Nov 11, 2026
67ddba5
phy: apple: t6040: split the USB2 sequences and the controller's rese…
fbalazsy Nov 11, 2026
1c29370
phy: apple: t6040: take a Type-C port's USB host through a deep sleep
fbalazsy Nov 11, 2026
f032724
usb: dwc3: apple: keep a T6040 Type-C port's host through a deep sleep
fbalazsy Nov 11, 2026
9a922a9
phy: apple: t6040: let a port keep its host through a sleep again aft…
fbalazsy Nov 11, 2026
c89e394
phy: apple: t6040: keep the host of a pin D port through a sleep whil…
fbalazsy Nov 11, 2026
bd442d1
phy: apple: t6040: drop the two CIO test switches
fbalazsy Nov 12, 2026
a32a1b5
usb: dwc3: apple: keep a Type-C port's host through a deep sleep with…
fbalazsy Nov 12, 2026
b40c25c
thunderbolt, usb: apple: say what the T6040 code does, not how it was…
fbalazsy Nov 12, 2026
2703d5d
Documentation: ABI: sn201202x: give the kernel version of hib_resume
fbalazsy Nov 12, 2026
13b1b9e
thunderbolt: apple: do not wait for the router stop in PM_POST_*
fbalazsy Nov 12, 2026
389e7d6
thunderbolt: apple: key the LSTX level at router stop on T6040
fbalazsy Nov 12, 2026
2d2ee87
thunderbolt: apple: keep the T6040 DP IN adapters disabled
fbalazsy Nov 12, 2026
f3af78b
usb: typec: tipd: give back what tipd_suspend() took after a cold boot
fbalazsy Nov 12, 2026
89ad7db
usb: typec: tipd: sn201202x: revalidate a port restored from disk itself
fbalazsy Nov 12, 2026
2679a73
usb: typec: tipd: sn201202x: poll the controller only while restoring it
fbalazsy Nov 12, 2026
f63f1e9
usb: typec: tipd: do not watch for a router whose node is disabled
fbalazsy Nov 12, 2026
e0670e4
thunderbolt: apple: owe a restart only to a port that had a cable
fbalazsy Nov 12, 2026
3893f91
usb: typec: tipd: sn201202x: do not wait for a wake interrupt that came
fbalazsy Nov 12, 2026
ca083b9
thunderbolt: apple: describe the T6040 router by what the hardware does
fbalazsy Nov 12, 2026
9992406
thunderbolt: apple: demote two expected T6040 messages to dev_dbg()
fbalazsy Nov 12, 2026
4b1d9c0
thunderbolt: apple: leave type7 where it was in apple_cio_hw
fbalazsy Nov 12, 2026
266469d
thunderbolt: apple: define apple_cio_start() below what it wraps
fbalazsy Nov 12, 2026
aa6a920
usb: typec: tipd: sn201202x: say what the restore from disk does now
fbalazsy Nov 12, 2026
5510534
usb: typec: tipd: sn201202x: use pm_sleep_ptr() for the sleep callbacks
fbalazsy Nov 12, 2026
de846d2
phy: apple: t6040: a powered-down sleep is the coprocessor's power cycle
fbalazsy Nov 12, 2026
14025bd
phy: apple: t6040: keep the host at USB2 when USB3 isn't back at a wake
fbalazsy Nov 12, 2026
cb7c353
phy: apple: t6040: try USB3 again at the next wake of a host left at …
fbalazsy Nov 12, 2026
35a2844
phy: apple: t6040: keep the host through sleeps again after a clean s…
fbalazsy Nov 12, 2026
38f4ce5
phy: apple: t6040: tolerate a busy recovery domain
RyanTheTide Nov 10, 2026
de1e9c6
phy: apple: t6040: don't wait for the PHY's own hold on the mailbox
fbalazsy Nov 12, 2026
d598aac
phy: apple: t6040: say which condition stops the host for a sleep
fbalazsy Nov 12, 2026
0af1674
phy: apple: t6040: shut the lanes down when the CIO bring-up fails
fbalazsy Nov 12, 2026
a5eb665
phy: apple: t6040: recover the coprocessor in the CIO bring-up too
fbalazsy Nov 12, 2026
408b38d
phy: apple: t6040: let DP wait for the host after an abandoned sleep
fbalazsy Nov 12, 2026
398da49
phy: apple: t6040: say who waits for the host's wake in DP's power-on
fbalazsy Nov 12, 2026
2779ef4
phy: apple: t6040: describe the PIPE and tunnel-clock steps by what t…
fbalazsy Nov 12, 2026
ce572f2
usb: dwc3: apple: don't report a GUID that wasn't read after a sleep
fbalazsy Nov 12, 2026
e33b305
usb: dwc3: apple: report a kept Type-C host's wake at the debug level
fbalazsy Nov 12, 2026
990f40f
phy: apple: t6040: drop the unused cio_flip
fbalazsy Nov 12, 2026
8a6ef15
phy: apple: t6040: give the tunable tables designated indices
fbalazsy Nov 12, 2026
87b5f64
phy: apple: t6040: one assignment per statement in the restore
fbalazsy Nov 12, 2026
4d767e9
phy: apple: t6040: one PHY shutdown for mode 2
fbalazsy Nov 12, 2026
a22d41e
phy: apple: t6040: a powered-down sleep ends a failed USB2 power-down
fbalazsy Nov 12, 2026
2b1b8de
phy: apple: t6040: keep the coprocessor powered through a suspend-to-…
fbalazsy Nov 12, 2026
f54115f
usb: dwc3: apple: suspend and resume asynchronously
fbalazsy Nov 13, 2026
9a95843
phy: apple: t6040: use one DPTX tunable table for USB3 and DP
fbalazsy Nov 13, 2026
29a98f4
usb: dwc3: apple: let the PHY decide on the PIPE reset after a sleep
fbalazsy Nov 13, 2026
bae65fe
phy: apple, soc: apple: give the DP tunnel pixel clock a symbol of it…
fbalazsy Nov 13, 2026
f5fbd1f
usb: typec: balance the module reference of an active altmode
fbalazsy Nov 13, 2026
1fb1d40
usb: dwc3: apple: decide a role during a sleep in one place
fbalazsy Nov 13, 2026
1197ec8
phy: apple: pick the PHY sleep provider by its ops
fbalazsy Nov 13, 2026
0998448
usb: typec: tipd: share the start of the application firmware
fbalazsy Nov 13, 2026
e000e5a
thunderbolt: apple: reach the Type-C PHY through struct phy
fbalazsy Nov 13, 2026
1ed2752
phy: apple, usb: typec: drop the restore callbacks' checks for an aba…
fbalazsy Nov 13, 2026
346a417
phy: apple: t6040: give DP IN 1 of a tunnel its PCLK1 prerequisite
fbalazsy Nov 13, 2026
5e66575
drm/apple: T6041 display coprocessors across hibernation and S2R
fbalazsy Nov 10, 2026
b63b475
drm/apple: iomfb: answer clear_default_surface on the 26.x desktop fi…
fbalazsy Nov 10, 2026
d53da2a
drm/apple: refuse a T6040 sleep the display firmware did not finish
fbalazsy Nov 10, 2026
f246627
drm/apple: report what the display got after a firmware restart
fbalazsy Nov 10, 2026
c197f44
drm/apple: log a CRTC mode set the firmware refused
fbalazsy Nov 10, 2026
1d880b9
drm/apple: pause and resume the DCP's AFK endpoints around a T6040 S2R
fbalazsy Nov 10, 2026
adb46a4
drm/apple: set the system power state on again after a T6040 S2R
fbalazsy Nov 10, 2026
ff0d014
drm/apple: send the panel level again once the firmware enables the b…
fbalazsy Nov 10, 2026
a0f874c
drm/apple: bring the external DCPs back after a T6040 S2R
fbalazsy Nov 11, 2026
5c3f6f3
drm/apple: let a crashed external display processor through a T6040 S2R
fbalazsy Nov 11, 2026
14bb1b5
drm/apple: give an external display's link back for a T6040 S2R
fbalazsy Nov 11, 2026
326656c
drm/apple: power-cycle the HDMI converter before a link replayed afte…
fbalazsy Nov 11, 2026
26e75f4
drm/apple: the HDMI converter's power follows its link
fbalazsy Nov 11, 2026
b5f5cf1
drm/apple: power a desktop 26.x external pipe on before a mode set on…
fbalazsy Nov 11, 2026
a3836b4
drm/apple: start the HDMI link of a desktop 26.x processor once per plug
fbalazsy Nov 11, 2026
aeb8a92
drm/apple: a display that fits one pipe again does not wait for a second
fbalazsy Nov 12, 2026
8fae0a7
drm/apple: keep the external displays on their pipes through a wake
fbalazsy Nov 12, 2026
9c4d2c2
drm/apple: name the T8140 driver instead of its pull request
RyanTheTide Nov 12, 2026
f44601b
drm/apple: stop a T6040 display coprocessor for a sleep in one way
fbalazsy Nov 12, 2026
ff33ca1
drm/apple: lower and raise a T6040 display coprocessor's CPU word in …
fbalazsy Nov 12, 2026
e242ed5
drm/apple: drop the progress marks of the T6040 sleep path
fbalazsy Nov 12, 2026
f852c1a
drm/apple: the cold session start is the resume from disk's only
fbalazsy Nov 12, 2026
f9c0f6b
soc: apple: sleep: t6040, drm: apple: say what the wake leaves undone
fbalazsy Nov 12, 2026
9f32574
drm/apple: dcp: include sleep.h and suspend.h once
fbalazsy Nov 12, 2026
c4e025a
drm/apple: dcp: drop the T6030 external-DCP tests from the T6041 slee…
fbalazsy Nov 12, 2026
a9d244f
drm/apple: dcp: register the restart notifier once, at probe
fbalazsy Nov 12, 2026
2ada041
drm/apple: dcp: build the firmware restart for every system sleep
fbalazsy Nov 12, 2026
a9286ee
drm/apple: dcp: name the power-state and GAPF fields of the T6041 sleep
fbalazsy Nov 12, 2026
a7ace96
drm/apple: dcp: describe the T6041 sleep by what the hardware does
fbalazsy Nov 12, 2026
665ef5b
drm/apple: dcp: log a T6040 sleep and wake at debug level
fbalazsy Nov 12, 2026
c81f328
drm/apple: dcp: say what the T6041 barriers order
fbalazsy Nov 12, 2026
ed490dd
mux: apple-display-crossbar: describe the T604X ATC registers plainly
fbalazsy Nov 12, 2026
d397ee6
drm/apple: dcp: keep a T6041 sleep to the pipe's own power words
fbalazsy Nov 12, 2026
7595f69
drm/apple: dcp: keep a given-up firmware stopped across a T6040 S2R
fbalazsy Nov 12, 2026
f09dc1d
drm/apple: dcp: enable the internal link work again after a T6040 S2R
fbalazsy Nov 12, 2026
5db5c4b
drm/apple: dcp: let a dead internal processor through a T6040 S2R
fbalazsy Nov 12, 2026
45f4503
drm/apple: dcp: count a T6041 processor out of pmc-dcp at its stop
fbalazsy Nov 12, 2026
590b949
drm/apple: dcp: drop two receive releases a T6040 S2R wake never needs
fbalazsy Nov 12, 2026
c7c834c
drm/apple: dcp: drop two calls that do nothing in the T6041 stop
fbalazsy Nov 12, 2026
021051f
drm/apple: dcp: keep the T6041 pipes' power words in the hw data
fbalazsy Nov 12, 2026
92062aa
drm/apple: dcp: map the T6041 sleep registers once, at probe
fbalazsy Nov 12, 2026
2ee07de
drm/apple: dcp: do not warn about silent AFK endpoints at a T6040 S2R
fbalazsy Nov 12, 2026
66cf0dd
drm/apple: leave out the mode set of an external pipe that has no sink
fbalazsy Nov 13, 2026
88b934f
drm/apple: dcp: state the restore kernel's DCP requirement
fbalazsy Nov 13, 2026
432f1ec
phy: apple: t6040: give the DP tunnel pixel clock a symbol of its own
fbalazsy Nov 13, 2026
4a960cd
drm/apple: dcp: share the firmware return of the S2R warm restarts
fbalazsy Nov 13, 2026
3a8bcc2
drm/apple: iomfb: drop includes the file does not use
fbalazsy Nov 13, 2026
584accf
drm/apple: report an idle that kept the firmware's memory apart from …
fbalazsy Nov 13, 2026
ca29f8e
tools: asahi: model the T6040 tunnel clock in the T6030 clock host test
fbalazsy Nov 13, 2026
d28fdad
drm/apple: dcp: state why the HDMI converter power check is per machine
fbalazsy Nov 13, 2026
59e2605
drm/apple: dcp: access the HDMI converter state flag without a data race
fbalazsy Nov 13, 2026
5b50e13
drm/apple: drop the restore callbacks' checks for an abandoned hibern…
fbalazsy Nov 13, 2026
7509ddf
drm/apple: dcp: start the HDMI link again after a sleep once the syst…
fbalazsy Nov 13, 2026
e07783f
drm/asahi: G17 GPU across hibernation and S2R on T6040
fbalazsy Nov 10, 2026
d60b4d9
drm/asahi: g17: find the GPU's PMGR word behind its PMP report entry
fbalazsy Nov 10, 2026
831b950
drm/asahi: g17: map the GPU's PMGR word for the commit stop as for th…
fbalazsy Nov 10, 2026
a30314a
drm/asahi: g17: keep the GPU's power word from before a commit stop
fbalazsy Nov 10, 2026
df544ce
drm/asahi: g17: take the T6040 GPU's power domain as its PMP report e…
fbalazsy Nov 10, 2026
170c40f
drm/asahi: drop the hibernation switches and the stop marks of the G1…
fbalazsy Nov 12, 2026
09ea0c4
drm/asahi: g17: lower the GPU's power word at the commit where no PMP…
fbalazsy Nov 12, 2026
ecc6ae6
drm/asahi: name the display notifiers the GPU drain runs ahead of
fbalazsy Nov 12, 2026
d03f430
drm/asahi: g17: say what the firmware does in the sleep comments
fbalazsy Nov 12, 2026
bd53abc
drm/asahi: g17: reset a kick ring after a cold firmware boot in one p…
fbalazsy Nov 13, 2026
4cb7607
drm/asahi: g17: keep the PMP handover wait with the GPU's other PMP code
fbalazsy Nov 13, 2026
2b05ef6
drm/asahi: move the GPU's PM notifiers out of the statistics shim
fbalazsy Nov 13, 2026
a9a2d7c
drm/asahi: g17: drop the second cache clean after a resume from disk
fbalazsy Nov 13, 2026
e2f1464
drm/asahi: g17: re-boot the firmware after hibernation off the PM not…
fbalazsy Nov 13, 2026
61fc99d
drm/asahi: g17: name the step that fails in a republish after hiberna…
fbalazsy Nov 14, 2026
a05f8d4
drm/asahi: g17: take the compute channel claim through the preparatio…
fbalazsy Nov 14, 2026
d601491
media: apple: isp: ask the sleep controller whether the power is lost
fbalazsy Nov 10, 2026
714a454
tools: asahi: restart the transport of a chip a sleep left without fi…
fbalazsy Nov 12, 2026
00216c5
tools: asahi: resume the core after a cold function's reset
fbalazsy Nov 12, 2026
c8bd51c
accel/ane: h16: call the sleeps that end on their timer timer_wakeups
fbalazsy Nov 12, 2026
d1cc3fa
accel/ane: h16: one predicate for the cold sleep cycle
fbalazsy Nov 12, 2026
e7093a5
media: apple: isp: ask the PM core whether a suspend takes the power
fbalazsy Nov 12, 2026
538fdd3
accel/ane: h16: describe the firmware, not where the facts came from
fbalazsy Nov 12, 2026
c871702
tools: asahi: hold no sleep for a restore the notifier refuses
fbalazsy Nov 12, 2026
b4ab02c
tools: asahi: have the alarm set again after a resume from disk
fbalazsy Nov 12, 2026
33de041
tools: asahi: learn a cold resume from the PM core, not from the PMP
fbalazsy Nov 13, 2026
43a4a16
tools: asahi: drop the restore callbacks' checks for an abandoned hib…
fbalazsy Nov 13, 2026
6ff6d32
accel/ane: h16: accept the provenance note of an HWX
fbalazsy Nov 13, 2026
3719799
accel/ane: h16: let __TEXT carry SG_NORELOC with or without a scratch
fbalazsy Nov 13, 2026
971566c
accel/ane: h16: skip the debugging entries of the HWX symbol table
fbalazsy Nov 13, 2026
e28dbaa
accel/ane: h16: size the HWX symbol table for whole models
fbalazsy Nov 13, 2026
5e6344d
accel/ane: h16: parse kernel data split over __KERN_0..__KERN_n
fbalazsy Nov 13, 2026
2526ff3
accel/ane: h16: take surface names longer than seven bytes
fbalazsy Nov 13, 2026
a3a9f4c
accel/ane: h16: bind split kernel data and constants of one program
fbalazsy Nov 13, 2026
2110d1a
accel/ane: h16: bound an H16 program by its total mapped size
fbalazsy Nov 13, 2026
0c3e185
selftests/ane: build H16 sections for split kernel data and constants
fbalazsy Nov 13, 2026
50e9dbd
Documentation/accel: describe the whole-model H16 HWX forms
fbalazsy Nov 13, 2026
90cf680
accel/ane: h16: send the declared buffer size when the section is sma…
fbalazsy Nov 13, 2026
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
25 changes: 25 additions & 0 deletions Documentation/ABI/testing/sysfs-driver-sn201202x
Original file line number Diff line number Diff line change
@@ -0,0 +1,25 @@
What: /sys/bus/spmi/devices/<dev>/hib_resume
Date: November 2026
KernelVersion: 7.4
Contact: František Balázsy <frantisekbalazsy@gmail.com>
Description:
(RW) How the last resume from hibernation left an Apple
SN201202x USB Type-C port controller. Reads one of:

none
No resume from hibernation since the last system
sleep, or "roles" was written since.
roles-pending
The controller was restored in place after the cold
boot behind the resume. The kernel revalidates the
port, without taking the port or a display on it
down, once the USB controller that provides its role
switch is bound.
failed
The in-place restore failed; the controller has to be
probed again (unbind and bind the device).

Writing "roles" while the state is "roles-pending" takes note
of it: the state then reads "none". The write is not needed
for the port to be revalidated. Other values fail with
-EINVAL, a write in another state with -EALREADY.
73 changes: 51 additions & 22 deletions Documentation/accel/apple-ane.rst
Original file line number Diff line number Diff line change
Expand Up @@ -139,7 +139,8 @@ Validated metadata retains whole-segment geometry and file extents, so one
TEXT allocation covers both task and constant sections. Scratch and external
surfaces have zero file bytes. The load planner reparses the authenticated
original instead of trusting cached metadata, rounds allocations to 16 KiB
and applies a 512 MiB aggregate host ceiling before any DMA allocation.
and applies a 2 GiB aggregate host ceiling (twice the segment limit) before
any DMA allocation.

The resource geometry helper requires one backend-owned mapping per planned
segment and one binding per external surface, resolved by the backend from
Expand Down Expand Up @@ -367,20 +368,32 @@ The validator consumes one kernel-owned immutable snapshot of the complete
HWX. It requires subtype 7 and bounded, exactly consumed load commands;
rejects unknown command kinds, duplicate descriptors, overlapping segments,
sections and file regions; and validates symbols against the complete
section-ordinal table. Tensor names, directions, normalized fp16 dimensions,
overflow-checked strides and full batched spans must agree with their
surfaces and buffer references. There must be at least one input and one
output, with the single output last.
section-ordinal table. Section symbol names are unique (checked in sorted
order: a whole model names tens of thousands of kernel blobs); STAB
entries, which describe source types, are skipped after a bound check of
their name. One compiler annotation (command 8) and one LC_NOTE provenance
note, whose payload lies inside the command after its 40-byte header, are
accepted and grant nothing. __TEXT may carry SG_NORELOC or no flags.
Tensor names, directions, normalized fp16 dimensions, overflow-checked
strides and full batched spans must agree with their surfaces and buffer
references; a reference holds a terminated name of up to 31 bytes after
+0x18 (32-byte commands for short names, 0x28 bytes for up to 15). There
must be at least one input and exactly one output.

Kernel data may be split over __KERN_0 to __KERN_15, in order, each with one
section __kern_n. Each further segment starts where the previous one ends in
VM and in the file, after a wholly file-backed one, so the segments form one
kernel region within the segment limit.

The linear and scratch program descriptors have exact fixed fields and
zero reserved bytes. All five 16-byte aperture records must equal whole
surface segment bases in order, followed by the optional kernel base;
reserved halves and unused records are zero. Interior, duplicate, reordered
and scratch/text aperture addresses are refused. Task counts and word counts
must agree with the independent stream walk. Relocations use the actual
kernel section ordinal, unique bounded offsets and valid addends, and may
patch only packed-record payload words. Headers, register selectors and
padding remain immutable.
zero reserved bytes. The 16-byte aperture records from +0x70 must equal the
whole surface segment bases in order, followed by every kernel section base
in order; reserved halves and unused records are zero. Interior, duplicate,
reordered and scratch/text aperture addresses are refused. Task counts and
word counts must agree with the independent stream walk. Each relocation
names one kernel section ordinal, uses a unique bounded offset and an addend
below that section's size, and may patch only packed-record payload words.
Headers, register selectors and padding remain immutable.

SHA-256 covers the entire original snapshot, including load commands and
padding. The snapshot digest is rechecked before execution. Validation
Expand Down Expand Up @@ -865,7 +878,19 @@ with the kernel slot index 1007. The uapi header names these indices.

Slot 1 carries the ``__KERN_0`` kernel section when the program has one and
otherwise the ``__TEXT,__const`` constants, which macOS loads in every
program.
program. Index 1007 in an operation pair names whatever slot 1 carries, so
BAR slot 1 (the constants) binds to it in a program without kernel data.

One slot gives one BAR base. A program whose kernel data are split over
several sections reads each through its own BAR, and a program with kernel
data and constant operands needs both bases. Such a program lists every
kernel section in the generic section after the surfaces and the scratch,
as a read-only input of the section size with its own ID, and binds each
kernel base to that ID; EXECUTE then carries one buffer per kernel section
at its offset in the kernel region mapping, and slot 1 carries the
constants. The driver takes this form when the generic section has the
kernel entries; the slot-1 kernel form accepts only one kernel section.
How the firmware treats the buffer form has not been run on hardware.

EXECUTE resources come from a kernel-owned map. The class follows the
resource usage mapping (inputs 0, the output 1). Each descriptor names its
Expand All @@ -887,14 +912,18 @@ ID, a malformed report or a timeout quarantines the session. Reports with
an unknown UUID never complete a request. Closing the file sends TERMINATE
then UNLOAD. Mappings stay until reboot.

One program lifetime per firmware boot is mandatory. On this firmware
family the first EXECUTE of a newly loaded program can write into the
previous program's output buffer when the buffer counts match, and a larger
output overruns that mapping. A second H16_LOAD therefore fails with -EBUSY
and the log line ``H16 one program per firmware boot; reboot required``,
also after the first program closed. The transport never publishes the
secure-mode commands 0x27, 0x28 and 0x34 or the exclave commands 0x30 and
0x31.
Several programs may be loaded in one firmware boot and executed in any
order. On this firmware family a direct (ISR) EXECUTE after a program change
can write into the previous program's output buffer when the buffer counts
match, and a larger output overruns that mapping. The first EXECUTE after
every change of program or process therefore carries request flag bit 0,
which the firmware's program manager (CAneProgramManagerHx) uses to refuse
the direct path and queue the request on the execution loop, as regular
traffic does. Program and process IDs are allocated by the firmware
from FIFO pools of 144 and 288 entries, so an UNLOADed ID is not reissued
until the pool wraps; the driver treats IDs as opaque. The transport never
publishes the secure-mode commands 0x27, 0x28 and 0x34 or the exclave
commands 0x30 and 0x31.

The ``ane-h16-program`` KUnit suite covers add-128 and synthetic slot
packing, LOAD/CREATE/TERMINATE/UNLOAD bodies and replies, the deny list,
Expand Down
2 changes: 2 additions & 0 deletions Documentation/devicetree/bindings/arm/apple/apple,pmgr.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -53,6 +53,8 @@ properties:
reg:
maxItems: 1

nonposted-mmio: true

"#address-cells":
const: 1

Expand Down
64 changes: 64 additions & 0 deletions Documentation/devicetree/bindings/chosen/aura,hibernate.yaml
Original file line number Diff line number Diff line change
@@ -0,0 +1,64 @@
# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
%YAML 1.2
---
$id: http://devicetree.org/schemas/chosen/aura,hibernate.yaml#
$schema: http://devicetree.org/meta-schemas/core.yaml#

title: Memory layout a boot object keeps the same on every boot

maintainers:
- František Balázsy <frantisekbalazsy@gmail.com>

description: |
A hibernation image puts every saved page back into the frame it was taken
from, so the boot that resumes an image has to give the kernel the same RAM,
and keep the same ranges from it, as the boot that wrote the image. On Apple
silicon the boot firmware loads the boot object at an address that changes
from boot to boot, and what the boot object keeps alive after the hand-off
moves with it. A boot object that instead keeps everything from the start of
RAM up to a fixed end from the kernel, at the same addresses on every boot,
says so with aura,hibernate-layout. Without it the kernel neither hibernates
nor resumes.

The boot firmware still places some firmware segments anew on every boot,
inside that range, and a kernel restored from an image holds the device tree
of the boot that wrote the image. The boot object therefore puts the device
tree of every boot at one address inside the range, given in
aura,hibernate-dt. The image does not contain that memory, so the restored
kernel reads there what the resuming boot placed.

The properties are named after aura, the boot loader that defined them and
whose released versions publish them.

select:
properties:
$nodename:
const: chosen
required:
- $nodename

properties:
$nodename:
const: chosen

aura,hibernate-layout:
$ref: /schemas/types.yaml#/definitions/uint64-array
description:
The range of RAM that is kept from the kernel and laid out the same on
every boot.
items:
- description: Physical address of the start of RAM.
- description: Physical address of the end of the range, exclusive.

aura,hibernate-dt:
$ref: /schemas/types.yaml#/definitions/uint64
description:
Physical address, inside the range of aura,hibernate-layout, of the
device tree the kernel is started with. The kernel reads at most 2 MiB
there.

dependencies:
aura,hibernate-dt: [ 'aura,hibernate-layout' ]

additionalProperties: true
...
48 changes: 48 additions & 0 deletions Documentation/devicetree/bindings/chosen/aurora,s2r-mailbox.yaml
Original file line number Diff line number Diff line change
@@ -0,0 +1,48 @@
# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
%YAML 1.2
---
$id: http://devicetree.org/schemas/chosen/aurora,s2r-mailbox.yaml#
$schema: http://devicetree.org/meta-schemas/core.yaml#

title: Return path of a boot object that owns the CPU reset vectors

maintainers:
- František Balázsy <frantisekbalazsy@gmail.com>

description: |
On Apple silicon the boot firmware locks every CPU's reset vector to the
boot object it started. A CPU that comes back from a power-off, and every
CPU after a suspend to RAM, therefore restarts in the boot object, not in the
kernel. A boot object that sends such a CPU back to the kernel says so with
this property: a secondary CPU returns through its spin-table release
address, and the boot CPU of a resume jumps, at EL2 with the MMU off, to the
address the kernel left in the mailbox. Its presence also tells the kernel
that its spin-table CPUs may leave the kernel and come back.

The mailbox is four 64-bit words in the CPU's byte order:
0: the magic value 0x3152325341525541 ("AURAS2R1"), written by the boot
object before the kernel starts
1: the MPIDR of the only CPU that may take the resume address
2: the resume address, set by the kernel before it suspends and cleared
by the boot object when a CPU takes it
3: how many times the resume address was taken

select:
properties:
$nodename:
const: chosen
required:
- $nodename

properties:
$nodename:
const: chosen

aurora,s2r-mailbox:
$ref: /schemas/types.yaml#/definitions/uint64
description:
Physical address of the mailbox, inside the boot object's own memory,
which the kernel does not use.

additionalProperties: true
...
Original file line number Diff line number Diff line change
@@ -0,0 +1,73 @@
# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
%YAML 1.2
---
$id: http://devicetree.org/schemas/display/apple,t6041-display-power.yaml#
$schema: http://devicetree.org/meta-schemas/core.yaml#

title: Apple T6040 display power words

maintainers:
- František Balázsy <frantisekbalazsy@gmail.com>

description: |
The words a T6040 (M4 Pro) display coprocessor's driver writes to lower
its power domains after it stops the firmware for a suspend-to-RAM and to
raise them again on the wake. A display coprocessor names this node in its
apple,display-power property. The words belong to the power manager and
the display block, so no driver binds it.

The internal display coprocessor has its power-state words in the SoC's
east power manager block, the external ones in the west block, each at its
own offsets; the display PMC word that counts the powered coprocessors is
shared by all three. The boot loader publishes one node for the internal
coprocessor and one that both external coprocessors name.

properties:
compatible:
const: apple,t6041-display-power

reg:
items:
- description: a register window, named by reg-names
- description: a register window, named by reg-names

reg-names:
description:
pmc-dcp is the display PMC word that counts the powered coprocessors,
ps the power manager block that holds the coprocessors' power-state
words. The two may come in either order.
oneOf:
- items:
- const: pmc-dcp
- const: ps
- items:
- const: ps
- const: pmc-dcp

required:
- compatible
- reg
- reg-names

additionalProperties: false

examples:
- |
soc {
#address-cells = <2>;
#size-cells = <2>;

display-power@5048a9c14 {
compatible = "apple,t6041-display-power";
reg = <0x5 0x048a9c14 0x0 0x8>,
<0x5 0x02280000 0x0 0x14000>;
reg-names = "pmc-dcp", "ps";
};

display-power@502800000 {
compatible = "apple,t6041-display-power";
reg = <0x5 0x02800000 0x0 0x14000>,
<0x5 0x048a9c14 0x0 0x8>;
reg-names = "ps", "pmc-dcp";
};
};
6 changes: 5 additions & 1 deletion Documentation/devicetree/bindings/media/apple,prores.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -248,8 +248,12 @@ allOf:
maxItems: 1
iommus:
maxItems: 1
resets:
description:
The shared codec reset. It also clears the
translations of the engine's DART, so a board may
leave it out and have a hung engine refused instead.
required:
- resets
- reg-names
- if:
properties:
Expand Down
2 changes: 1 addition & 1 deletion Documentation/devicetree/bindings/pci/apple,pcie.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -504,7 +504,7 @@ allOf:
properties:
reg:
minItems: 8
maxItems: 8
maxItems: 9
interrupts:
minItems: 6
maxItems: 6
Expand Down
28 changes: 28 additions & 0 deletions Documentation/devicetree/bindings/phy/apple,t6040-atcphy.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -172,6 +172,29 @@ properties:
$ref: "#/$defs/apple,tunable"
description: USB2 host tunables, applied to the bank0 region.

apple,tunable-cio-lane0:
$ref: "#/$defs/apple,tunable"
description: >
Lane 0 receiver defaults for CIO (USB4/Thunderbolt) mode, applied to the
core region when the port enters USB4 or Thunderbolt. CIO mode is offered
only when both lane tables are present.

apple,tunable-cio-lane1:
$ref: "#/$defs/apple,tunable"
description: Lane 1 receiver defaults for CIO mode, applied to the core region.

apple,tunable-cio-uc-regs:
$ref: "#/$defs/apple,tunable"
description: Optional PHY microcontroller defaults for CIO mode, core region.

apple,tunable-cio-lane0-eq:
$ref: "#/$defs/apple,tunable"
description: Optional lane 0 receiver equalization for CIO mode, core region.

apple,tunable-cio-lane1-eq:
$ref: "#/$defs/apple,tunable"
description: Optional lane 1 receiver equalization for CIO mode, core region.

required:
- compatible
- reg
Expand Down Expand Up @@ -199,6 +222,11 @@ allOf:
mode-switch: false
port: false
ports: false
apple,tunable-cio-lane0: false
apple,tunable-cio-lane1: false
apple,tunable-cio-uc-regs: false
apple,tunable-cio-lane0-eq: false
apple,tunable-cio-lane1-eq: false

additionalProperties: false

Expand Down
Loading
Loading