I didn't notice anything wrong when I added my Roland device IDs to your patch's method. Its information can be found here, but I really doubt it applies to your issue, other than Takashi Iwai's simplified capture-quirk-only table patch (which Geraldo already mentioned): https://bugzilla.kernel.org/show_bug.cgi?id=212519 Since I built my patch upon your previous patch, is there anything I can do to test it on my devices for you?