The qcom-wled OVP interrupt is enabled at probe whenever the module is already on, but the driver still believes the backlight is off, so the first brightness update re-enables the interrupt and trips "Unbalanced enable for IRQ". sdm845-mainline has carried a workaround for years and Joel posted a flag-based version last October [1], which Konrad and Daniel asked to be replaced by something that does not track the IRQ state by hand.
Patch 1 is the drvdata fix I said I would drop after v3. The objection was that it stored the wrong object, not that the fix was wrong, so it is back with the right one stored. Patch 2 is unrelated too and fell out of writing the readback: the WLED3 brightness register stride has been one byte instead of two since the 2019 restructuring, so multi-string pm8941 boards only program one string correctly. Untested, I have no WLED3 board; it restores what pm8941-wled.c did and matches the downstream leds-qpnp.c layout. [1] https://lore.kernel.org/all/20251021-qcom-wled-fix-unbalanced-ovp-irq-enable-v2-1-7ff115b4f...@joelselvaraj.com/ Signed-off-by: David Heidelberg <[email protected]> --- Changes in v4: - I applied the 0x10 fix too late, now it's fixed in the right commit. (Konrad) - Right object stored in the drvdata. (Konrad) - Set the drvdata next to the other wled fields instead of next to the backlight device registration. - Reworded the drvdata commit, the NULL dereference is in wled_remove(), the driver has no PM callbacks. - Link to v3: https://patch.msgid.link/[email protected] Changes in v3: - Fix typo 0x10 != 0b10. (Sashiko) - Added fix for a unrelated issue, but since reported as critical, let's squeeze it: Missing platform_set_drvdata() in wled_probe() leads to a guaranteed NULL pointer dereference in wled_remove(). (Sashiko) - Link to v2: https://patch.msgid.link/[email protected] Changes in v2: - Add a fix for the WLED3 brightness register stride. - Add patch reading the programmed brightness back via backlight_ops.get_brightness. (Konrad) - Note the brightness readback is done in the follow-up to the first patch. - Link to v1: https://patch.msgid.link/[email protected] --- David Heidelberg (4): backlight: qcom-wled: Fix NULL pointer dereference in wled_remove() backlight: qcom-wled: Fix WLED3 brightness register stride backlight: qcom-wled: Fix unbalanced OVP IRQ enable at probe backlight: qcom-wled: Read back the programmed brightness at probe drivers/video/backlight/qcom-wled.c | 89 +++++++++++++++++++++++++++++++------ 1 file changed, 76 insertions(+), 13 deletions(-) --- base-commit: 944a035ecca915ae947905dcfb03f2b9dc6d032c change-id: 20260908-qcom-wled-backlight-fd9574027353 Best regards, -- David Heidelberg <[email protected]>
