Linux-mediatek Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Erik Steinkamp <erik@donoper.de>
To: linux-mediatek@lists.infradead.org
Subject: linux-firmware-mediatek 1:20260810-1 breaks Bluetooth keyboard input
Date: Thu, 27 Aug 2026 16:12:19 +0200	[thread overview]
Message-ID: <ca2a108e-ab07-4ebc-bb51-9851d173b63f@donoper.de> (raw)

Dear all,


I've observed something that might be worth for you to know:


Summary:

     Upgrading linux-firmware-mediatek from 1:20260622-1 to 1:20260810-1 
breaks Bluetooth keyboard input (BLE HID-over-GATT) on a MediaTek 
Bluetooth controller. Downgrading back to 20260622-1 fixes it 
immediately, with no other package changes.

Hardware:
     - Bluetooth controller: MediaTek MT7925
     - Peripheral: Keychron K2 HE keyboard, BLE HID-over-GATT (HOGP), 
current QMK firmware
     - Host: CachyOS, kernel linux-cachyos 7.1.8-1 (also reproduced on 
7.2.0-1), bluez 5.87

Symptom:

     the keyboard pairs and bluetoothctl reports Connected: yes, but no 
keystrokes ever register — no input device is created at all (checked 
via evtest/libinput).

bluetoothd logs:

     bluetoothd: src/service.c:service_accept() input-hog profile accept 
failed for <Keyboard-MAC>


A btmon capture shows the actual timing:

     HCI Command: LE Start Encryption ... #64 13.258318
     HCI Event: Command Status (Success) ... #65 13.258704
     bluetoothd: input-hog profile accept failed ... 13.260434
     HCI Event: Encryption Change (Success, AES-CCM) ... #66 13.303230


The HID-over-GATT setup in bluetoothd (a known no-retry race, see 
bluez/bluez#1911) gives up ~43ms before the controller actually reports 
Encryption Change. With linux-firmware-mediatek 20260622-1, the 
controller apparently completes and reports encryption fast enough to 
win this race; with 20260810-1, it consistently doesn't, so every BLE 
HID reconnect fails.

Steps to reproduce:
     1. Pair a BLE HID (HOGP) keyboard/mouse to a MediaTek Bluetooth 
controller running firmware 20260810-1.
     2. Observe it connects but produces no input, with the input-hog 
error above.
     3. Downgrade to linux-firmware-mediatek 20260622-1, rebuild 
initramfs, reboot.
     4. Reconnect the same device — works correctly every time.

Let me know if you need anything else.

Cheers,

Erik



                 reply	other threads:[~2026-08-27 14:12 UTC|newest]

Thread overview: [no followups] expand[flat|nested]  mbox.gz  Atom feed

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=ca2a108e-ab07-4ebc-bb51-9851d173b63f@donoper.de \
    --to=erik@donoper.de \
    --cc=linux-mediatek@lists.infradead.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox