[PATCH wireless-next 0/2] wifi: mt76: mt7925: LED support for USB adapters

[Josef Schlehofer]

MT7925 USB adapters such as the NETGEAR A9000 have a status LED on
GPIO 8 of the chip. MediaTek's Windows driver asks the firmware to
switch it with UNI command 0x2a, but the firmware in linux-firmware
does not act on that command [1].

This series drives the pin from the host instead. The layout of the
CBTOP GPIO block comes from the coda headers in MediaTek's gen4m driver
for MT6639, which uses the same CBTOP register map as MT7925. The first
patch lets the mt76 core register an LED that only provides
brightness_set_blocking, which a USB driver needs. The second patch
adds the LED to mt7925.

Tested on a Turris Omnia with an A9000 and firmware 20231227 over
several days. The LED follows the throughput trigger and manual
brightness changes, and turns off when the radio goes down.

Sean, JB, is it fine to drive these GPIO registers from the host, or
does the firmware rely on GPIO 8 staying in its default function?

[1] https://lore.kernel.org/linux-wireless/20261012125310.13917-1-pepe.schlehofer@xxxxxxxxx/

Josef Schlehofer (2):
  wifi: mt76: allow LEDs with only brightness_set_blocking
  wifi: mt76: mt7925: add LED support for USB adapters

 drivers/net/wireless/mediatek/mt76/mac80211.c | 12 ++++-
 .../net/wireless/mediatek/mt76/mt7925/init.c  | 51 +++++++++++++++++++
 .../net/wireless/mediatek/mt76/mt7925/mcu.c   |  2 +
 .../net/wireless/mediatek/mt76/mt7925/regs.h  |  8 +++
 drivers/net/wireless/mediatek/mt76/mt792x.h   |  2 +
 5 files changed, 73 insertions(+), 2 deletions(-)


base-commit: 991a73f4909f1348ffb4b38765cdb522a775e4c2
-- 
2.54.0 (Apple Git-157)