Skip to content

leds: pwm: do not force the LED off when period readback is zero - #219

Merged
chriskearney merged 1 commit into
omacom:integration/aurorafrom
ludagoo:fix/leds-pwm-zero-period
Oct 30, 2026
Merged

chriskearney merged 1 commit into
omacom:integration/aurorafrom
ludagoo:fix/leds-pwm-zero-period

Conversation

@ludagoo

@ludagoo ludagoo commented Oct 25, 2026 •

Copy link
Copy Markdown

leds-pwm treated a zero PWM period readback, with default-state "keep", as a failed read and then applied duty 0. On Apple FPWM that readback is the reset value (ON_CYCLES and OFF_CYCLES both zero). The period from the pwms cell is already in pwm->args.

When the hardware period is 0 and the args period is not, keep that period for later brightness updates and leave the output unchanged. Apply off only when there is still no period to use.

This is the keyboard backlight on a MacBook Air M1 2020 (J313): fpwm1 channel 0, pwms period 40000 ns, default-state = "keep". Probe was logging "failed to read period for kbd_backlight, default to off" and forcing the LED off before systemd-backlight restored the saved level.

The fpwm1 suspend power-state failure is a separate bug and is not changed here.

Session build, Monday 26 Oct 2026 05:02 PM -03: linux-aurora 7.2.11-6-2-ARCH (package 7.2.11.aurora6-2), branch local/j313-test at 7638df0. makepkg exited 0. rustc 1.85.0, LLVM/clang 18.1.8. This machine is running that kernel. The stacked commit compiled in was 066a421. This pull request's head is 7cd67b0, which is not the commit that was built.

The new probe message appears at boot. malik-na confirmed the keyboard backlight lit early in boot and stayed lit.

@chriskearney chriskearney left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Reviewed 70c122ce2677885e8ae3d44aa1fe84b7fcd55fa2. The fallback obtains period/polarity from the PWM arguments while retaining KEEP, so probe still skips the call that would apply zero duty. I found no source blocker in that change. The reported brightness can remain zero until userspace writes a value; that differs from turning the hardware output off.

Native ARM64 package build passes for this head and the base control with the same recipe/toolchain. I independently checked the as-built pin and packaged config: CONFIG_LEDS_PWM=y, so the changed driver is compiled in.

No J313 boot test yet. The remaining hardware check is output retention through probe followed by normal brightness control; suspend restoration is a separate change in #221.

@chriskearney

Copy link
Copy Markdown

Tested 70c122ce2677885e8ae3d44aa1fe84b7fcd55fa2 on an M1 MacBook Air (J313), using a native aarch64 build with CONFIG_LEDS_PWM=y (7.2.11-6-219.1-ARCH). One-shot boot, desktop smoke test and keyboard LED sysfs checks passed. Brightness accepted/read back 1 and 63 of 255, then restored its original value of 0.

The tester reported 38 pass, 9 fail and 26 skip, with no functional regressions against the installed-kernel reference. Two additional lab-only failures are the expected installed-package/boot-chain checks for a temporary kernel. The remaining failures were already present. This verifies the brightness interface and readback; it does not measure physical light output or demonstrate a previously failing probe condition. No suspend/resume test ran.

The machine returned to its installed 7.2.11-6-1-ARCH kernel. Protected boot/kernel files and tester files match their pre-test hashes; Wi-Fi, internal/external display operation and GL/Vulkan pixel checks passed after restoration, with no new matched GPU/DART/kernel faults. Receipt: job j1025-172103-d46f; kernel artifact SHA256 1f45ed8686c794a3943267c994568af8fa6108c17f99eef694473a174b7342ca.

@malik-na

Copy link
Copy Markdown
Member

Tested on a J313 (M1 Air): works, no regressions found.

baseline (19f4965180f4) this PR
failed to read period for kbd_backlight, default to off at probe 1 0
kbd_backlight brightness after boot 219 219
set 128 → read back 128 128
after the s2idle cycle, set 200 → read back 200 200

All lab tests passed, including the device suspend phase and omarchy-m-test with no regressions against the installed kernel. The PS fpwm1: Failed to reach power state 0xf (… now: 0xa4f) after the s2idle cycle is unchanged, which is expected: that's #221's bug.

How this was tested
  • Hardware: MacBook Air (M1, 2020), J313, BCM4378 (brcmfmac4378b1-pcie), DCP 13.5, booted through U-Boot → Limine.
  • Kernels: this PR's head and its parent 19f4965180f4 (the baseline), built with the Air's own /proc/config.gz and booted once each as a one-shot test entry. The Mac returns to its installed 7.2.11-6-1-ARCH after every run.
  • Each boot ran:
    • a device-phase suspend test (pm_test=devices through systemctl suspend)
    • omarchy-m-test, compared with a run on the installed kernel
    • a check script that reads dmesg counters, writes the keyboard and panel backlight (420, 419, 189, then back), lists ALS IIO channels, re-associates Wi-Fi twice with nmcli con up, and does one pm_test=platform s2idle cycle (noirq + resume_early, auto-wakes after 5 s)
  • Not covered: a real s2idle with a keypress wake, Touch ID enrollment/verify with a finger, and how brightness looks.

@malik-na

Copy link
Copy Markdown
Member

Follow-up: confirmed by eye on the Mac. Watching the keyboard during boot, the backlight went dark only while the Mac was powered off. It lit early in boot and didn't go dark again; dmesg shows no PWM period read back for kbd_backlight; using 40000 ns and leaving the output unchanged. On every baseline boot (4 of 4) the same probe logs failed to read period … default to off.

How this was tested
  • MacBook Air (M1, 2020), J313, DCP 13.5, BCM4378, booted through U-Boot → Limine.
  • This PR's head and its parent 19f4965180f4 (the baseline), built with the Air's own /proc/config.gz and booted as one-shot test entries. The Mac returns to its installed 7.2.11-6-1-ARCH after every run.
  • Automated boots: device-phase suspend (pm_test=devices), omarchy-m-test, and a check script (dmesg counters, keyboard and panel backlight writes, ALS channels, two nmcli con up re-associations, one pm_test=platform s2idle cycle).
  • Hands-on boots: a person at the Mac doing a real lid-close sleep and keypress wake, using the brightness keys, and covering or lighting the ALS, while the kernel's state was logged every 2 s.
  • Tested commit: 70c122ce2677 on 19f4965180f4. The PR has since been rebased to 7cd67b0b4db4 on 937b727ef9a3 with an identical diff; the new base wasn't re-tested.

@ludagoo
ludagoo marked this pull request as ready for review October 26, 2026 22:58
@ludagoo
ludagoo force-pushed the fix/leds-pwm-zero-period branch from 7cd67b0 to f3c796e Compare October 28, 2026 15:32
Apple FPWM comes out of reset with ON_CYCLES and OFF_CYCLES both zero.
pwm-apple get_state reports that as period 0. leds-pwm, on
default-state = "keep", treats a zero period as a failed read, switches
to off, and applies duty 0.

On J313 the keyboard backlight is that device: fpwm1 channel 0,
pwms = <&fpwm1 0 40000>, default-state = "keep". Every boot logs

  leds_pwm led-controller: failed to read period for kbd_backlight, default to off

and probe programs a 0% waveform before systemd-backlight can restore
the saved level.

The period in the pwms cell is already in pwm->args by the time
led_pwm_add() runs. When the hardware period is 0 and that args period
is not, keep the args period for later brightness updates and do not
apply a duty of 0. Apply off only when there is still no period to use.

The fpwm1 suspend power-state failure is a separate bug and is not
touched here.

Signed-off-by: Lucas Goossen <lucas.goossen@hey.com>
@ludagoo
ludagoo force-pushed the fix/leds-pwm-zero-period branch from f3c796e to 23e938c Compare October 30, 2026 10:14

@chriskearney chriskearney left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Source approval for 23e938c. The one-file change applies cleanly to integration/aurora 12be047 and reproduces driver blob 3865cd7bee86c74048f38f42b5ae73f98a38c163.

On the KEEP path with zero readback period, pwm_init_state prepares local metadata from the PWM arguments; it does not apply hardware state. A usable argument period retains KEEP and skips the existing probe-time PWM apply. Nonzero readback, explicit ON/OFF and missing-period fallback remain unchanged. Later brightness, active-low, suspend and error handling remain on their existing paths. The optional enable-GPIO path can still use reported brightness zero; that is preexisting and the J313 DT has no such GPIO, so this is not a universal claim that every board's output controls are untouched.

Independent exact-function/helper host controls passed 15 cases, with a failing pre-change KEEP control and three caught semantic mutants. I verified the report/checksums and read the changed driver/helper path. These mocks are not hardware execution. The driver bytes match the older hardware-observed version, but those observations do not prove a boot of the entire current rebased kernel.

Required Kernel CI run 41039245386 was held for fork approval and is now running after normal approval. No current-head CI pass is claimed here. Queueing/merge must wait for the required result and normal full combined-tree merge-queue checks; no CI or recovery bypass.

@chriskearney
chriskearney added this pull request to the merge queue Oct 30, 2026
Merged via the queue into omacom:integration/aurora with commit 3225ab3 Oct 30, 2026
14 checks passed
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.

3 participants