Re: [PATCH v5 2/4] ASoC: codecs: Add TAS67524 quad-channel audio amplifier driver

[Mark Brown]

On Fri, Apr 10, 2026 at 12:56:47PM -0500, Sen Wang wrote:
> On 4/10/26 09:02, Mark Brown wrote:

> > This looks mostly good, but one issue I see is that AFAICT we only stop
> > fault_check_work during runtime suspsend - if runtime PM is disabled, or
> > if the driver is removed, the work will be left running.

> Regarding fault_check_work w/o runtime PM: the device has fault
> conditions that fire regardless of stream state, so the work runs for the
> duration of the active power state (runtime PM) rather than stream state.

> I could add a DAPM event fallback to gate the work when runtime PM is not
> defined, but that's more restrictive; it could miss faults outside of active
> playback, and adds complexity for what is a fairly niche configuration
> (optional check_work with runtime PM disabled).

> Do you think the DAPM fallback would suffice, or is the current approach
> (poll until removal) acceptable given the hardware behavior? Any other
> suggestions would be greatly appreciated!

It's fine to keep on checking for faults if there's faults that can be
generated, the only reason I mentioned runtime PM there was that it's
the only thing that stops the polling in the current version.  So long
as everything is stopped when the device is removed it's fine.  No need
for a DAPM fallback.

Attachment: signature.asc
Description: PGP signature