Skip to content

arm64: dts: exynos: google: zuma-husky: haptics and the rear torch - #200

Merged
derekross merged 5 commits into
derekross:omamobilefrom
Somnius:omamobile-zuma-husky-haptics-torch
Nov 7, 2026
Merged

derekross merged 5 commits into
derekross:omamobilefrom
Somnius:omamobile-zuma-husky-haptics-torch

Conversation

@Somnius

@Somnius Somnius commented Nov 4, 2026 •

Copy link
Copy Markdown
Contributor

Haptics and the rear torch on the Pixel 8 Pro. Both work on my phone.

Everything comes from Google's GPL trees (zuma-shusky-haptics.dtsi, zuma-husky-camera.dtsi, zuma-usi.dtsi, cmucal-sfr.c) and follows what the 7 Pro does, with the same two drivers that are already in the tree (cs40l26-vibra and leds-lm3644).

4 commits:

  1. dt-bindings: the clock IDs for PERIC1's USI0, the PERIC0 paths in CMU_TOP and the new CMU_PERIC0 with USI14, plus google,zuma-cmu-peric0 and google,zuma-peric0-sysreg.
  2. clk-zuma: all the offsets below.
  3. Haptics: CS40L26A at 0x43 on PERIC1 USI0 (hsi2c_0, IRQ 656; the clocks are USI0 mux 0x620, divider 0x1808 and gates 0x2038, 0x203c, 0x208c). zuma.dtsi gets usi0, hsi2c_0 and the HSI1 pin controller (pinctrl_hsi1, IRQ 413, gph0 to gph2, alias pinctrl8). The board file has the node with the stock boost and actuator values (2.5 A, 11 V, 145 Hz, 13 ohm), INTb gpa6-3, reset gph2-3.
  4. Torch: LM3644 at 0x63 on PERIC0 USI14 (hsi2c_14, IRQ 648). This needs a new CMU_PERIC0 (0x10800000) and the PERIC0 paths in CMU_TOP: TOP mux 0x10f0/0x10ec, gates 0x2114/0x2110, dividers 0x18e8/0x18e4; PERIC0 user muxes 0x600/0x640, divider 0x180c, gates 0x2068/0x206c/0x20cc, read-only GPIO and sysreg gates 0x20b8/0x20f0. HWEN is gpp8-2. The torch is capped at 50 mA per LED, same as the 7 Pro's bring-up limit, not a stock number.

All offsets are from Google's zuma cmucal-sfr.c, and they match the PERIC1 ones that are already here.

Tested on my 8 Pro, 7.3.0-rc6 (the e1 kernel, so Brett's tag plus my own commits, not plain omamobile), with cs40l26.wmfw and cs40l26.bin from my own phone's vendor image in /lib/firmware (not in the repo):

  • haptics: CS40L26 ID 0x40a260, revision 0xa1, Firmware 0x1800d4 rev 7.2.91 running, cs40l26_vibra input device; a 3-pulse FF_RUMBLE test buzzed hard.
  • torch: LM3644TT (ID 0x04), 17 steps (2550 to 49110 uA), white:torch lights the LED at step 4.
  • Wi-Fi and Bluetooth still fine next to both.

On current omamobile the series applies and clk-zuma and zuma-husky.dtb build, but I did not boot that exact tree and I did not run dt_binding_check / dtbs_check.

One thing for you to decide, @derekross @greenbrettmichael: the base DTS kept CMU_PERIC1 and its sysreg off and only the battery bring-up overlays turned them on. Haptics needs them, so this enables cmu_peric1, sysreg_peric1, cmu_peric0 and sysreg_peric0 in zuma-husky.dts on purpose. If you'd rather keep them off in the board file and use an overlay, say so and I'll move them.

Not tested: haptics through feedbackd (it has no udev rule on the 8 Pro yet), anything above the 50 mA torch cap, suspend. The shell's torch tile is not tried yet either.

@derekross

Copy link
Copy Markdown
Owner

Thank you, @Somnius, for haptics and the torch together, and for following the 7 Pro's nodes so closely. That
makes this easy to read. I checked every offset against Google's cmucal-sfr.c and cmucal-node.c: the CMU_TOP
PERIC0 paths, the new CMU_PERIC0 and PERIC1's USI0, with their parents and divider widths. They're all right. So
are the interrupts, the sysreg offsets and the HSI1 controller against zuma-usi.dtsi and stock's zuma.dtsi, and
the CS40L26A values against zuma-shusky-haptics.dtsi. Keeping the GPIO and sysreg gates read-only is exactly
right. Our gate passes on today's omamobile, also merged with #199.

Before it goes in:

  1. Clock IDs. @greenbrettmichael's GPU PR (Pixel 8 Pro GPU: zuma's CMU_G3D, ACPM GPU clock, panthor and a GPU thermal zone #157, and Pixel 8 Pro CPU DVFS: zuma's CPU levels from the ECT and CPU thermal zones (stacked on #157) #158 on top of it) adds CMU_TOP IDs 22-34 to the same
    block where this one adds 22-27. Git merges that block without a conflict and keeps both, so we'd get
    duplicate IDs. I'll tell you both which one goes first. The second renumbers its TOP IDs after the other's, with
    nr_clk_ids to match.
  2. The HSI1 pin controller is enabled in zuma.dtsi, so every zuma board probes it, the Pixel 8a included,
    and gpiolib reads its registers at probe. It clearly works on your 8 Pro, but HSI1 is the PCIe block, and on
    gs101 reading a pin controller Linux doesn't own has aborted the SoC. Could you make it status = "disabled"
    in zuma.dtsi and enable it in zuma-husky.dts?
  3. Two comments. In zuma.dtsi, the CMU_PERIC0 block landed directly under the existing "CMU_PERIC1 and USI11
    …" comment, which now describes the wrong node. Could you move the PERIC0 nodes above that comment? And the
    torch block in zuma-husky.dts says CMU_PERIC0 is enabled "for the test; a plain build keeps them off". Since
    this is the board file, the plain build does enable them, so it just needs rewording.
  4. The test and the tree that lands, as on arm64: dts: exynos: google: zuma-husky: Bluetooth on the APM UART #199: the uname -r, the blob ids of the touched files in the e1
    tree, where its tag lives, and a tools/phone-smoke report. For the haptics, the probe lines and
    cat /sys/class/input/<the cs40l26 device>/capabilities/ff would round it off.

On your question about enabling CMU_PERIC1 and CMU_PERIC0 in the board file: that's fine with me. Their GPIO and
sysreg gates are read-only, and the 8 Pro boots with clk_ignore_unused, so nothing unused gets switched off.
Keep it in the board file.

Two optional things: stock turns on the CS40L26's brownout protection at 3.008 V
(cirrus,vpbr-threshold-microvolt = <3008000> in our binding), which might be worth having with a 2.5 A boost.
And upstream will want the sysreg compatible as its own dt-bindings: soc: samsung: exynos-sysreg: commit.

This one will need a small rebase after #199 (both add to samsung,exynos-sysreg.yaml and the end of
zuma-husky.dts: keep both sides). Thanks again!

@Somnius

Somnius commented Nov 6, 2026

Copy link
Copy Markdown
Contributor Author

@derekross thank you. Reworked on today's omamobile (29a7a93), on my fork as zuma-husky-haptics-torch-v2 for now (https://github.com/Somnius/linux-omarchy-mobile/tree/zuma-husky-haptics-torch-v2). I'll move this PR's branch onto it after #199, and I'll say here when I do, since it rewrites the history.

What changed: the pinctrl_hsi1 node is status = "disabled" in zuma.dtsi and zuma-husky.dts enables it, with a comment saying why. The CMU_PERIC0 nodes sit above the "CMU_PERIC1 and USI11" comment now, and the torch comment says the plain build does enable them. I added cirrus,vpbr-threshold-microvolt = <3008000> from stock, the sysreg compatible is its own dt-bindings: soc: samsung: exynos-sysreg: commit, and the haptics commit message says the pin controller is disabled in the dtsi.

tools/kernel-pr-check passes on the 5 commits (checkpatch --strict clean, W=1, device trees build), and dt-validate on the merged husky dtb prints only the two eusb2 phy supply messages that omamobile already has.

The clock IDs: I'm not touching them until you say whether Brett's #157 and #158 or this one goes first; the second one renumbers its TOP IDs after the other's with nr_clk_ids to match. For the test info (uname -r, blob ids, tag, phone-smoke, the cs40l26 probe lines and capabilities/ff) I'm waiting for a boot of the combined #199 and #200 tree on omamobile, same as on #199.

@Somnius
Somnius force-pushed the omamobile-zuma-husky-haptics-torch branch from fbd6bc7 to 09d4f92 Compare November 6, 2026 05:07
@Somnius

Somnius commented Nov 6, 2026

Copy link
Copy Markdown
Contributor Author

I force-pushed this branch (now 09d4f92) with the reworked series from my last comment: sysreg compatible split out, pinctrl_hsi1 disabled in the dtsi and enabled in the board, PERIC0 nodes moved, vpbr threshold added. Boot test of the combined tree with #199 is still to come.

@Somnius

Somnius commented Nov 6, 2026

Copy link
Copy Markdown
Contributor Author

I booted the combined tree (origin/omamobile + #199 + #200, merge b0fb5bc, tag husky-test-20261106-pr199-200 on my fork) on my 8 Pro. uname -r 7.3.0-rc6-omamobile-push1-gb0fb5bccb3d6, it was up 223 s when I ran tools/phone-smoke and all the required checks passed.

Blob ids at that commit:

  • arch/arm64/boot/dts/exynos/google/zuma-husky.dts b3fcd8716155f6fdc877cffe4e386ed2b33fb6b8
  • arch/arm64/boot/dts/exynos/google/zuma.dtsi 21c184ae870ca6cf83466456dac5a28089b76228
  • arch/arm64/boot/dts/exynos/google/zuma-pinctrl.dtsi 74b69056313b2dbb602e513df5fe86f1e7227106
  • drivers/clk/samsung/clk-zuma.c 833d676e8b92cf5e600604a7db87d65ed39b94dc

Haptics/torch side: cs40l26 probes (Cirrus Logic CS40L26 ID 0x40a260, revision 0xa1, cs40l26.wmfw loads, supplies vdd-a and vdd-p fall back to dummy regulators), input device cs40l26_vibra exists and capabilities/ff reads 107030000 0. lm3644 probes too (torch: 17 steps, 2550-49110 uA per channel) and /sys/class/leds/white:torch is there. I did not play a vibration effect or light the torch in this boot, so those two are probe-only so far.

@derekross

Copy link
Copy Markdown
Owner

Thank you, @Somnius, for turning every point around so fast. HSI1 is now off in zuma.dtsi and on in the board file, the PERIC0 nodes sit above the right comment, the torch comment says what the file does, and the brownout threshold and the sysreg split are in. I fetched husky-test-20261106-pr199-200, which is exactly today's omamobile with #199 and this PR, and the blob ids match. The cs40l26 and lm3644 probe lines and the capabilities/ff value look as they should (that value is plain FF_RUMBLE, the same as the 7 Pro reports).

On the clock IDs: let's have this PR go first. #157 and #158 need a rebase anyway, and I'll ask Brett there to start his CMU_TOP IDs at 28. So you don't need to renumber anything.

What's left before it goes in:

  1. Once arm64: dts: exynos: google: zuma-husky: Bluetooth on the APM UART #199 is merged, please rebase onto omamobile. The conflicts are the ones you already resolved in your test tree: keep both compatibles in samsung,exynos-sysreg.yaml (apm, then peric0) and both blocks at the end of zuma-husky.dts (Bluetooth, then haptics). arm64: dts: exynos: google: zuma-husky: NFC controller node (disabled) #208 and clk: samsung: exynos-clkout: NFC clock pad on the Pixel 8 Pro (zuma) #209 then rebase onto the new head.
  2. Please paste the tools/phone-smoke report from that boot into the thread.

A tiny wording point, optional: the comment above &pinctrl_hsi1 says the board enables it "for the haptics' interrupt and reset lines", but the interrupt is gpa6-3, on an alive bank. HSI1 carries the reset and trigger lines. Also, #205 enables cmu_peric1 and sysreg_peric1 too, so whichever of the two lands second can drop its copy (the duplicates are harmless either way).

Thanks again, this is a lovely pair of features.

@Somnius

Somnius commented Nov 7, 2026

Copy link
Copy Markdown
Contributor Author

@derekross here is the same tools/phone-smoke report as on #199, from the combined tree boot of husky-test-20261106-pr199-200 (kernel 7.3.0-rc6-omamobile-push1-gb0fb5bccb3d6, up 223 s). It was recovered from my session log, not re-run. The GPU and the two local units (omamobile-sound, omamobile-thermal) read "not ok" on that tree, details on #199.

Kernel 7.3.0-rc6-omamobile-push1-gb0fb5bccb3d6, tools at 722e530, 2026-11-06T05:12Z.

Check Result Detail
boot (required) ok kernel 7.3.0-rc6-omamobile-push1-gb0fb5bccb3d6, slot b, up 223s
storage guard (required) ok 62 partitions; only userdata and slot B's boot images writable
root (required) ok ext4 on userdata, 206G free
kernel health (required) ok no oops or BUG; 0 WARN, 18 error lines in dmesg
systemd units not ok failed: omamobile-sound.service omamobile-thermal.service
USB gadget (required) ok bound to 11210000.usb
display ok card0-Unknown-1 connected, 1344x2992
GPU not ok no render node
session (required) ok Hyprland and the Omarchy shell running
lock ok PAM file present, the shell can lock (passwordPam true, locked true)
touch ok SALIXP Touchscreen
buttons ok gpio-keys
Wi-Fi skip no wireless interface
Bluetooth ok hci0, powered true
battery ok 92% Charging, 33 °C
sound cards skip none
sensors skip no sensor proxy
cameras skip no libcamera tools
thermal ok hottest zone 33 °C

I'll rebase onto omamobile once #199 is in, keeping both compatibles in samsung,exynos-sysreg.yaml (apm, then peric0) and both blocks at the end of zuma-husky.dts (Bluetooth, then haptics). Since #205 is merged now, I'll drop my copy of cmu_peric1 and sysreg_peric1 in the same rebase.

@derekross

Copy link
Copy Markdown
Owner

Thank you, @Somnius. The report covers it, and your rebase plan is right: both compatibles in samsung,exynos-sysreg.yaml (apm, then peric0), both blocks at the end of zuma-husky.dts (Bluetooth, then haptics), and dropping your cmu_peric1 and sysreg_peric1 enables now that #205 has them. I've merged #199, so go ahead whenever you're ready. Once CI is green on the rebased head, I'll merge this, and then #208 and #209 can rebase on top.

Add the clock IDs for PERIC1's USI0 (the haptics driver's I2C bus on
the Pixel 8 Pro), the PERIC0 NoC and peripheral paths in CMU_TOP, and
the new CMU_PERIC0 with USI14 (the rear flash LED's I2C bus), and the
google,zuma-cmu-peric0 compatible.

Signed-off-by: Lef <me@lefteros.com>
The sysreg of the Tensor G3's PERIC0 block, which holds the USI mode
selectors; the Pixel 8 Pro's rear flash LED bus (USI14) needs it.

Signed-off-by: Lef <me@lefteros.com>
PERIC1's USI0: user mux 0x620, divider 0x1808, gates 0x2038, 0x203c and
0x208c. CMU_TOP's PERIC0 paths: muxes 0x10f0 and 0x10ec, gates 0x2114
and 0x2110, dividers 0x18e8 and 0x18e4. CMU_PERIC0 at 0x10800000: NoC
user mux 0x600, USI14 user mux 0x640, USI14 divider 0x180c, USI14 gates
0x2068, 0x206c and 0x20cc, and the read-only GPIO and sysreg gates
0x20b8 and 0x20f0.

All offsets are from Google's GPL zuma cal-if (cmucal-sfr.c) and match
the PERIC1 ones that are already here. Tested on a Pixel 8 Pro:
haptics (USI0) and the torch (USI14) both probe on these clocks.

Signed-off-by: Lef <me@lefteros.com>
Add usi0 and hsi2c_0 (0x10c80000, IRQ 656) and the HSI1 pin controller
(pinctrl_hsi1 at 0x12040000, IRQ 413, gph0 to gph2, alias pinctrl8) to
zuma.dtsi; the pin controller is disabled there and the board enables
it, as HSI1 is the PCIe block and every zuma board would otherwise
probe it. The board file gets the CS40L26A at 0x43 with the stock
boost, actuator and brownout values from Google's
zuma-shusky-haptics.dtsi (2.5 A, 11 V, 145 Hz, 13 ohm, 3.008 V), INTb
on gpa6-3 and RESET on gph2-3.

USI0 sits on CMU_PERIC1 and SYSREG_PERIC1, which the board file enables
since the hsi2c_11 change.

Tested on a Pixel 8 Pro with 7.3.0-rc6: the chip answers (ID 0x40a260,
revision 0xa1), firmware 7.2.91 runs from cs40l26.wmfw and cs40l26.bin
(taken from the owner's own image, not in the repo), and a
force-feedback rumble effect vibrates the phone.

Signed-off-by: Lef <me@lefteros.com>
Add cmu_peric0, sysreg_peric0, usi14 and hsi2c_14 (0x10980000, IRQ 648)
to zuma.dtsi, and the pins hsi2c14_bus (gpp16-0 and gpp16-1, function
2) and flash_en (gpp8-2). The board file gets the LM3644 at 0x63 with
HWEN on gpp8-2 and the 7 Pro's bring-up cap of 50 mA per LED (not a
stock value), and enables cmu_peric0 and sysreg_peric0, which stay off
in zuma.dtsi like PERIC1.

Tested on a Pixel 8 Pro with 7.3.0-rc6: the chip is found (LM3644TT,
ID 0x04), the LED class white:torch has 17 steps from 2550 to 49110 uA
and lights the rear LED.

Signed-off-by: Lef <me@lefteros.com>
@Somnius
Somnius force-pushed the omamobile-zuma-husky-haptics-torch branch from 09d4f92 to 25079bb Compare November 7, 2026 16:34
@Somnius

Somnius commented Nov 7, 2026

Copy link
Copy Markdown
Contributor Author

@derekross rebased onto today's omamobile and force-pushed, because a rebase rewrites the five commits (same five, new head 25079bb on 5a0e057). Done as you said: both compatibles in samsung,exynos-sysreg.yaml (apm, then peric0), both blocks at the end of zuma-husky.dts (Bluetooth, then haptics), and I dropped my cmu_peric1 and sysreg_peric1 enables since #205 has them; the haptics commit message now says the board file enables them since the hsi2c_11 change. I also fixed the comment above &pinctrl_hsi1: HSI1 carries the reset and trigger lines, and the interrupt, gpa6-3, is on an alive bank.

git range-diff against my old branch shows commits 3 and 5 unchanged, and 1 and 2 differing only in context from #199. The dtb builds with W=1 and checkpatch --strict is clean on all five. I haven't booted this head: the smoke report on the thread is from the earlier test tag. CI is green on the new head.

@derekross
derekross merged commit 25368bf into derekross:omamobile Nov 7, 2026
1 check passed
@derekross

Copy link
Copy Markdown
Owner

Merged, thank you @Somnius. #208 and #209 can rebase onto omamobile now; as I said there, I'll merge #209 (with #208's commit in it) once the messages are updated and it's green.

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.

2 participants