Skip to content

HID: dockchannel-hid: allow models without stm to proceed - #648

Closed
yuyuyureka wants to merge 36 commits into
AsahiLinux:asahi-wipfrom
yuyuyureka:feature/stm
Closed

yuyuyureka wants to merge 36 commits into
AsahiLinux:asahi-wipfrom
yuyuyureka:feature/stm

Conversation

@yuyuyureka

@yuyuyureka yuyuyureka commented Oct 14, 2026 •

Copy link
Copy Markdown

The MacBook Neo does not have an stm, and the current code indefinitely waits for the stm report to come in to retrieve the keyboard and touchpad serials. Fix this by only waiting for the stm, if the stm subnode is present in the device tree (which is already the case in the M2-M3 device trees).


Please check all boxes that apply:

  • This code was developed in accordance with the Asahi Linux Copyright Policy
  • I have not used generative AI/LLMs to develop this code, in accordance with the Asahi Linux LLM Policy
  • I have certified the above by adding a Signed-off-by: trailer to every commit

All boxes must be checked before your pull request can be reviewed.

jannau and others added 30 commits October 14, 2026 18:13
Certain Broadcom bluetooth chips (bcm4377/bcm4378/bcm438) need ACL
streams carrying audio to be set as "high priority" using a vendor
specific command to prevent 10-ish second-long dropouts whenever
something does a device scan. This patch sends the command when the
socket priority is set to TC_PRIO_INTERACTIVE, as BlueZ does for audio.

Signed-off-by: Sasha Finkelstein <fnkl.kernel@gmail.com>
The current approach of silently disabling all rust drivers if the
toolchain is missing results in users that try to compile their own
kernels getting a "successful" build and then being confused about where
did their drivers go. In comparison, missing openssl results in a build
failure, not a disappearance of everything that depends on it.

This also means that allyesconfig will depend on rust, but since the
rust experiment concluded with "rust is here to stay", i believe that
allyesconfig should be building rust drivers too.

Signed-off-by: Sasha Finkelstein <k@chaosmail.tech>
Apple M3 Pro and Max devices are using 'gp00' keys for GPIO in addition
to 'gP00' keys. Add a second compatible to handle this keys with an
additional macsmc-gpio instance.

Signed-off-by: Janne Grunau <j@jannau.net>
Add support for SMC GPIO keys with a lower letter 'p' via the
"apple,smc-low-gpio" compatible. This adds support for a second
macsmc-gpio controller using 'gp00' keys.
These keys are used on Apple M3 Pro and Max MacBooks in the controller
for keyboard and trackpad and for the built-in DisplayPort to HDMI
converter.

Signed-off-by: Janne Grunau <j@jannau.net>
Apple M3 Pro and Max devices are using 'gp00' keys for GPIO in addition
to 'gP00' keys. These keys are handled by an additional macsmc-gpio
instance using the "apple,smc-low-gpio" compatible.

Signed-off-by: Janne Grunau <j@jannau.net>
Signed-off-by: Hojun Cho <ghwns6743@gmail.com>
Signed-off-by: Janne Grunau <j@jannau.net>
sofus13 and others added 6 commits October 14, 2026 19:06
Signed-off-by: sofus <sofus.c@icloud.com>
Some userspace applications use this to know what capture node to open.

Signed-off-by: sofus <sofus.c@icloud.com>
Besides the interface creation rework, this commit also dropped the
initialisation of roam_delta in brcmf_dongle_roam(). Since then
WLC_SET_ROAM_DELTA is sent with uninitialised stack data, which the
BCM4387 firmware rejects:

  ieee80211 phy0: brcmf_dongle_roam: WLC_SET_ROAM_DELTA error (-52)

Restore the two lines to match mainline (20 dB delta on all bands).

Signed-off-by: Matthias Kurz <m.kurz@irregular.at>
Signed-off-by: Hojun Cho <ghwns6743@gmail.com>
Signed-off-by: Sofus Forstreuter <sofus.c@icloud.com>
The MacBook Neo does not have an stm, and the current code indefinitely
waits for the stm report to come in to retrieve the keyboard and
touchpad serials. Fix this by only waiting for the stm, if the stm
subnode is present in the device tree (which is already the case in the
M2-M3 device trees).

Co-authored-by: Michael Reeves <michael.reeves077@gmail.com>
Signed-off-by: Michael Reeves <michael.reeves077@gmail.com>
Signed-off-by: Yureka Lilian <yureka@cyberchaos.dev>
@jannau

jannau commented Oct 14, 2026

Copy link
Copy Markdown
Member

I guess this is ok for the downstream driver as this will be replaced with the upstream version soon. Is that already considered there? I really need to get around reviewing that.

For #573 we will need a way to identify haptic touchpads (with stm) anyway.

@IntegralPilot

IntegralPilot commented Oct 14, 2026 •

Copy link
Copy Markdown

Upstream NAKed an empty child node for stm and instead want apple,has-stm on parent (which I think sounds good and will include in upstream v5 which I'm sending after 7.4-rc1).

But this might be fine for downstream for now, I don't really mind.

@jannau

jannau commented Oct 15, 2026

Copy link
Copy Markdown
Member

merged manually, thanks

@jannau jannau closed this Oct 15, 2026
hitori-chan added a commit to hitori-chan/linux that referenced this pull request Nov 19, 2026
The core powers down in a WFI with CYC_OVRD's WFI mode (bits 25:24) at 3
and comes back with its system registers, SP and PC, losing x0-x30 and
the FP/SIMD registers. It is XNU's one idle WFI (arm64_retention_wfi):
macOS idles every core in it whenever its next timer is 50 us or more
away. CPU PD reaches the same retention only past a 10 ms prediction and
behind every CPU PM notifier.

RET keeps x18-x30 on the stack around the WFI and saves and releases the
FP state in place of cpu_pm_enter(). No other notifier that can run on
this SoC needs it: arm_pmu's counts survive, arch_timer's CNTKCTL is
unchanged, a KVM guest ran clean, and MPAM and CoreSight are absent.
Every other WFI keeps m1n1's mode 2 (clock-gated).

Measured on AsahiLinux#648 and AsahiLinux#656, 2026-11-18 and 19:
- ret-q: 41,573 of 41,597 entries retained (x19 zeroed), d8 lost on the
  same ones; cntkctl, vbar, tpidr, mdcr_el2, pmcr0, pmcr1 and fpcr never
  changed; no FP pattern lost in 80,000 sleeps.
- ret3-q, PSTR, the mean of two cycles after a warm-up (spread 0.01-0.09
  W): video 11.635 -> 11.465 W, scroll 13.645 -> 13.475 W. CPU PD at
  500 us saves as much, but wakes about 3 us later.
- retres-q: a residency of 50, 200 or 500 us saves the same within
  0.15 W. A wake costs about 11 us over a WFI (a 500 us sleep's p50:
  16.4 and 15.3 us in RET, 4.9 and 4.7 us in WFI), hence 12 us.
- ret-gd-q: Grim Dawn +0.28 fps (stdev 0.29), +0.008 J a frame.
- ret-wifi4-q: 24 of 24 suspends, the gateway back 2-3 s after each
  wake, no Wi-Fi firmware error; 1.9 % of the pings lost a cycle, 0.8 %
  over 12 cycles with RET off.
hitori-chan added a commit to hitori-chan/linux that referenced this pull request Nov 20, 2026
The core powers down in a WFI with CYC_OVRD's WFI mode (bits 25:24) at 3
and comes back with its system registers, SP and PC, losing x0-x30 and
the FP/SIMD registers. It is XNU's one idle WFI (arm64_retention_wfi):
macOS idles every core in it whenever its next timer is 50 us or more
away. CPU PD reaches the same retention only past a 10 ms prediction and
behind every CPU PM notifier.

RET keeps x18-x30 on the stack around the WFI and saves and releases the
FP state in place of cpu_pm_enter(). No other notifier that can run on
this SoC needs it: arm_pmu's counts survive, arch_timer's CNTKCTL is
unchanged, a KVM guest ran clean, and MPAM and CoreSight are absent.
Every other WFI keeps m1n1's mode 2 (clock-gated).

Measured on AsahiLinux#648 and AsahiLinux#656, 2026-11-18 and 19:
- ret-q: 41,573 of 41,597 entries retained (x19 zeroed), d8 lost on the
  same ones; cntkctl, vbar, tpidr, mdcr_el2, pmcr0, pmcr1 and fpcr never
  changed; no FP pattern lost in 80,000 sleeps.
- ret3-q, PSTR, the mean of two cycles after a warm-up (spread 0.01-0.09
  W): video 11.635 -> 11.465 W, scroll 13.645 -> 13.475 W. CPU PD at
  500 us saves as much, but wakes about 3 us later.
- retres-q: a residency of 50, 200 or 500 us saves the same within
  0.15 W. A wake costs about 11 us over a WFI (a 500 us sleep's p50:
  16.4 and 15.3 us in RET, 4.9 and 4.7 us in WFI), hence 12 us.
- ret-gd-q: Grim Dawn +0.28 fps (stdev 0.29), +0.008 J a frame.
- ret-wifi4-q: 24 of 24 suspends, the gateway back 2-3 s after each
  wake, no Wi-Fi firmware error; 1.9 % of the pings lost a cycle, 0.8 %
  over 12 cycles with RET off.
hitori-chan added a commit to hitori-chan/linux that referenced this pull request Nov 20, 2026
The core powers down in a WFI with CYC_OVRD's WFI mode (bits 25:24) at 3
and comes back with its system registers, SP and PC, losing x0-x30 and
the FP/SIMD registers. It is XNU's one idle WFI (arm64_retention_wfi):
macOS idles every core in it whenever its next timer is 50 us or more
away. CPU PD reaches the same retention only past a 10 ms prediction and
behind every CPU PM notifier.

RET keeps x18-x30 on the stack around the WFI and saves and releases the
FP state in place of cpu_pm_enter(). No other notifier that can run on
this SoC needs it: arm_pmu's counts survive, arch_timer's CNTKCTL is
unchanged, a KVM guest ran clean, and MPAM and CoreSight are absent.
Every other WFI keeps m1n1's mode 2 (clock-gated).

Measured on AsahiLinux#648 and AsahiLinux#656, 2026-11-18 and 19:
- ret-q: 41,573 of 41,597 entries retained (x19 zeroed), d8 lost on the
  same ones; cntkctl, vbar, tpidr, mdcr_el2, pmcr0, pmcr1 and fpcr never
  changed; no FP pattern lost in 80,000 sleeps.
- ret3-q, PSTR, the mean of two cycles after a warm-up (spread 0.01-0.09
  W): video 11.635 -> 11.465 W, scroll 13.645 -> 13.475 W. CPU PD at
  500 us saves as much, but wakes about 3 us later.
- retres-q: a residency of 50, 200 or 500 us saves the same within
  0.15 W. A wake costs about 11 us over a WFI (a 500 us sleep's p50:
  16.4 and 15.3 us in RET, 4.9 and 4.7 us in WFI), hence 12 us.
- ret-gd-q: Grim Dawn +0.28 fps (stdev 0.29), +0.008 J a frame.
- ret-wifi4-q: 24 of 24 suspends, the gateway back 2-3 s after each
  wake, no Wi-Fi firmware error; 1.9 % of the pings lost a cycle, 0.8 %
  over 12 cycles with RET off.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

6 participants