On 12/6/2024 1:09 AM, Cezary Rojewski wrote: > On 2024-12-04 12:15 AM, Wesley Cheng wrote: >> >> On 12/3/2024 8:13 AM, Cezary Rojewski wrote: >>> On 2024-11-06 8:34 PM, Wesley Cheng wrote: >>>> In order to allow userspace/applications know about USB offloading status, >>>> expose a sound kcontrol that fetches information about which sound card >>>> and PCM index the USB device is mapped to for supporting offloading. In >>>> the USB audio offloading framework, the ASoC BE DAI link is the entity >>>> responsible for registering to the SOC USB layer. > > ... > >>> R) += mixer_usb_offload.o >>>> diff --git a/sound/usb/mixer_usb_offload.c b/sound/usb/mixer_usb_offload.c >>>> new file mode 100644 >>>> index 000000000000..e0689a3b9b86 >>>> --- /dev/null >>>> +++ b/sound/usb/mixer_usb_offload.c >>>> @@ -0,0 +1,102 @@ >>>> +// SPDX-License-Identifier: GPL-2.0 >>>> +/* >>>> + * Copyright (c) 2022-2024 Qualcomm Innovation Center, Inc. All rights reserved. >>>> + */ >>>> + >>>> +#include <linux/usb.h> >>>> + >>>> +#include <sound/core.h> >>>> +#include <sound/control.h> >>>> +#include <sound/soc-usb.h> >>> >>> ALSA-components should not be dependent on ASoC ones. It should be done the other way around: ALSA <- ASoC. >>> >> >> At least for this kcontrol, we need to know the status of the ASoC state, so that we can communicate the proper path to userspace. If the ASoC path is not probed or ready, then this module isn't blocked. It will just communicate that there isn't a valid offload path. > > I'm not asking _why_ you need soc-usb.h header, your reasoning is probably perfectly fine. The code hierarchy is not though. If a sound module is dependent on soc-xxx.h i.e. ASoC symbols, it shall be part of sound/soc/ space. I'm still reviewing the HDAudio flow a bit, so please correct me if I'm wrong. During module initialization, it looks like there will be some overall platform card that will call snd_hdac_ext_bus_init() to create the HDA bus. I referred to the Intel AVS core. How do you ensure that the ALSA entities are loaded before this call goes out? I think once the bus is created dynamic creation/removal of HDA devices is fine, and the hdev_attach/detach is executed. Thanks Wesley Cheng
