Re: [RFC PATCH v6 0/2] the Topping M62's vendor controls, on the component framework

[Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com>]

Hi Takashi,

A gentle ping on this. The problem is unchanged: on Linux the M62's
preamp gains, output volumes and source selectors can only be set by
hand on the front panel. v6 is the component split you suggested; v8
stays posted as the alternative.

Since posting, v6 has run on the hardware as sent: all nine controls,
panel and ALSA following each other, fifty module unload/load cycles
and the cable pulled out mid-write three times, under KASAN and
lockdep. Nothing in the log.

The one wart in the cover letter is worse than it says. Reloading the
HID module takes the controls off a live card and puts them back, and
ALSA reports that correctly, but WirePlumber does not merely lose its
faders -- it segfaults. That is for WirePlumber to fix and an issue is
open there, but it is a cost of this road that v8 does not have.

One thing points the other way. The card also reports its battery
level unasked, and the natural home for that is a power_supply with
device scope, as hid-corsair-void does. In a HID driver that is a
small patch on top; on the v8 road it would have to be registered
from a snd-usb-audio mixer quirk, and I cannot find anything there
that does so. I have not written it and would like to know the road
first.

So the question is the one from the cover letter: this road or v8?

Jiri, Benjamin: the needs_remote_wakeup question is still open --
whether clearing it after hid_hw_open() is acceptable, or whether
usbhid should offer a way to take input reports without it.

Thanks,
Mikhail