Re: [PATCH] Input: atkbd - skip deactivate for Lenovo IdeaPad Slim 3 14IWC11

From: Martino Papero

Date: Tue Sep 29 2026 - 16:37:26 EST



Il 19/09/26 02:09, Martino Papero ha scritto:

Il 11/09/26 20:52, Hans de Goede ha scritto:
Hi,
I'm starting to wonder if we should not just make this behavior the default.

IIRC the whole deactivate thing was added as a workaround for multi-byte scancodes
getting split-up during suspend/resume. I wonder if we can come-up with some
better fix for that ?

Regards,

Hans


Hi,

I tested the same atkbd_deactivate_fixup on a Lenovo IdeaPad Slim 3 15IWC11 (83RR), which was affected by the keyboard issue previously discussed here:

https://lore.kernel.org/all/20260628-ideapad-kbd-leds-v1-0-50b421d94927@xxxxxxxxx/ <https://lore.kernel.org/all/20260628-ideapad-kbd-leds-v1-0-50b421d94927@xxxxxxxxx/>

Without any kernel quirk, the internal keyboard does not work. I had been using:

|i8042.nopnp=1 i8042.dumbkbd=1|

which makes the keyboard work, but prevents the Caps Lock LED from working.

I built Fedora kernel 7.2.5 with the following additional DMI entry:

{
    /* Lenovo IdeaPad Slim 3 15IWC11 (83RR) */
    .matches = {
    DMI_MATCH(DMI_SYS_VENDOR, "LENOVO"),
    DMI_MATCH(DMI_PRODUCT_NAME, "83RR"),
    },
    .callback = atkbd_deactivate_fixup,
},|

I booted this kernel without any i8042 command-line parameters.

The internal keyboard and Caps Lock LED both work correctly. I also tested suspend/resume and a cold boot, with no issues so far.

DMI:
LENOVO 83RR
LENOVO_MT_83RR_BU_idea_FM_IdeaPad Slim 3 15IWC11

So it looks like the 83RR can use the same atkbd_deactivate_fixup as the 83RQ.

I can test a revised patch if needed.

Regards,
Martino


Hi,

Just following up on this.

The 83RR is still working reliably with atkbd_deactivate_fixup after further use, without i8042.nopnp or i8042.dumbkbd.

Would it be useful for me to submit a separate patch adding the Lenovo IdeaPad Slim 3 15IWC11 (83RR) to the atkbd_deactivate_fixup DMI table?

I can prepare and test the patch on the affected hardware. Alternatively, if the generic solution discussed above is preferred, I am happy to test that as well.

Regards,
Martino