Re: wifi fixes that might be good to have in 7.2 final

[Devin Wittmayer]

On 12/08/2026 13:57, Thorsten Leemhuis wrote:
> P.S.: I wonder if "wifi: mt76: mt7921: refactor regd update to fix
> recursive mutex deadlock") (which is as 1f83a8378eb26c in net-next [...]
> ) would be good to have in 7.2, too.

It is the same bug, and that patch fixes it. I ran it A/B on an MT7922
with mt7921e, swapping only mt7921-common.ko: stock deadlocks on a 6 GHz
association, patched associates at 5975 MHz in four seconds with no splats.

The fix is in wireless-next, not in wireless.git. v7.2-rc7 still takes
dev->mutex unconditionally in mt7921_mcu_regd_update(), so 7.2 ships the
deadlock unless it moves trees.

Nothing is owed to stable: e88098133ed4 is in v7.2-rc1 and not in v7.1.

Devin