Attachment 6343 Details for Bug 19950 – Патч в апстрим v1

[patch] Патч в апстрим v1

positron-atkbd-fix.patch (text/plain), 4.55 KB, created by Mikhail Novosyolov on 2026-02-18 07:19:47 MSK

(hide)

Description:

Filename:

MIME Type:

Creator: Mikhail Novosyolov

Created: 2026-02-18 07:19:47 MSK

Size: 4.55 KB

patch

obsolete

>From f120fd4985692bee2462ff1588c7c327f0d2257d Mon Sep 17 00:00:00 2001
>From: Mikhail Novosyolov <m.novosyolov@rosa.ru>
>Date: Wed, 18 Feb 2026 06:52:48 +0300
>Subject: [PATCH] Input: atkbd: add keymap fixup for notebooks using 0x6e as Fn
> modifier
>
>Commit dc8c9c171ef3 ("Input: atkbd - map F23 key to support default
>copilot shortcut") mapped scancode 0x6e to KEY_F23 to support the
>Microsoft Copilot key on Lenovo, HP, and Dell notebooks.
>
>However, some notebook platforms (including Positron Proxima 15 and
>possibly others based on the same OEM design) use scancode 0x6e for the
>Fn modifier key instead of a dedicated Copilot key. When 0x6e generates
>KEY_F23 events, the Fn key breaks Fn combinations such as Fn+F5
>(touchpad toggle).
>
>On these platforms, the hardware relies on 0x6e being unmapped to
>properly handle Fn combinations at the firmware level. When the kernel
>maps it to KEY_F23, desktop environments intercept this as a global
>hotkey and toggle the touchpad, but cannot re-enable it because the
>firmware no longer recognizes Fn as a valid modifier.
>
>Userspace solutions (systemd hwdb) cannot fix this because the keycode
>mapping happens in the atkbd driver before events reach userspace.
>A kernel-level quirk is required.
>
>Add a DMI-based keymap fixup to remap scancode 0x6e to 0 for affected
>systems. Currently only Positron Proxima 15 is known to be affected,
>but other notebooks based on the same OEM platform may exhibit the same
>behavior and can be added to the quirk table.
>
>Scancode 0x6e has different meanings on different hardware:
>- Lenovo/HP/Dell: F23 (Copilot key) - correctly mapped to KEY_F23
>- Some OEM platforms: Fn modifier - must be unmapped
>
>A generic solution is not feasible without breaking Copilot support
>on other vendors.
>
>Link: https://linux-hardware.org/?probe=7aca7ed668
>Link: https://bugzilla.rosa.ru/show_bug.cgi?id=19950
>
>Fixes: dc8c9c171ef3 ("Input: atkbd - map F23 key to support default copilot shortcut")
>Co-developed-by: GLM-4.7 AI <noreply@z.ai>
>Signed-off-by: Mikhail Novosyolov <m.novosyolov@rosa.ru>
>---
> drivers/input/keyboard/atkbd.c | 51 ++++++++++++++++++++++++++++++++++
> 1 file changed, 51 insertions(+)
>
>diff --git a/drivers/input/keyboard/atkbd.c b/drivers/input/keyboard/atkbd.c
>index 6c999d89ee4b..e4ecdd9fc07b 100644
>--- a/drivers/input/keyboard/atkbd.c
>+++ b/drivers/input/keyboard/atkbd.c
>@@ -1068,6 +1068,40 @@ static unsigned int atkbd_volume_forced_release_keys[] = {
> 	0xae, 0xb0, -1U
> };
> 
>+/*
>+ * Positron notebooks where scancode 0x6e is used by the Fn key and should
>+ * not generate KEY_F23 events.
>+ *
>+ * Commit dc8c9c171ef3 ("Input: atkbd - map F23 key to support default
>+ * copilot shortcut") mapped scancode 0x6e to KEY_F23 for Copilot support.
>+ * However, on Positron notebooks scancode 0x6e is generated by the Fn
>+ * modifier key and should be unmapped to allow firmware/hardware to
>+ * process Fn+F5 and other Fn combinations properly.
>+ */
>+static unsigned int atkbd_positron_fn_keymap_fixup_scancodes[] = {
>+	0x6e, -1U
>+};
>+
>+/*
>+ * Fixup to remap scancodes to 0 (no keycode) for machines where
>+ * the default keymap incorrectly assigns keys that should be unused.
>+ * This is needed when BIOS/firmware uses certain scancodes for
>+ * internal purposes (e.g., Fn modifier) and they should not generate
>+ * keyboard events.
>+ */
>+static void atkbd_apply_keymap_fixup(struct atkbd *atkbd, const void *data)
>+{
>+	const unsigned int *scancodes = data;
>+	unsigned int i;
>+
>+	for (i = 0; scancodes[i] != -1U; i++) {
>+		unsigned int scancode = scancodes[i];
>+
>+		if (scancode < ATKBD_KEYMAP_SIZE)
>+			atkbd->keycode[scancode] = 0;
>+	}
>+}
>+
> /*
>  * OQO 01+ multimedia keys (64--66) generate e0 6x upon release whereas
>  * they should be generating e4-e6 (0x80 | code).
>@@ -1775,6 +1809,14 @@ static int __init atkbd_setup_forced_release(const struct dmi_system_id *id)
> 	return 1;
> }
> 
>+static int __init atkbd_setup_keymap_fixup(const struct dmi_system_id *id)
>+{
>+	atkbd_platform_fixup = atkbd_apply_keymap_fixup;
>+	atkbd_platform_fixup_data = id->driver_data;
>+
>+	return 1;
>+}
>+
> static int __init atkbd_setup_scancode_fixup(const struct dmi_system_id *id)
> {
> 	atkbd_platform_scancode_fixup = id->driver_data;
>@@ -1937,6 +1979,15 @@ static const struct dmi_system_id atkbd_dmi_quirk_table[] __initconst = {
> 		},
> 		.callback = atkbd_deactivate_fixup,
> 	},
>+	{
>+		/* Positron Proxima 15 - Fn key (0x6e) should not generate F23 */
>+		.matches = {
>+			DMI_MATCH(DMI_SYS_VENDOR, "LNPO Positron LLC"),
>+			DMI_MATCH(DMI_PRODUCT_NAME, "G1569"),
>+		},
>+		.callback = atkbd_setup_keymap_fixup,
>+		.driver_data = atkbd_positron_fn_keymap_fixup_scancodes,
>+	},
> 	{ }
> };
> 
>-- 
>2.51.0
>

Actions: View | Diff

Attachments on bug 19950: 6343