On Tue, Jun 16, 2026 at 6:48 PM Mark Brown <[email protected]> wrote: > > On Tue, Jun 16, 2026 at 11:00:38AM +0800, [email protected] wrote: > > From: Shengjiu Wang <[email protected]> > > > > After pm_runtime_enable() the AUDMIX hardware block is powered off and > > remains suspended until the first runtime resume. Any register write > > issued between probe() and the first resume (for example from DAPM or > > ALSA control paths) would target unpowered hardware and be silently > > lost. > > > platform_set_drvdata(pdev, priv); > > pm_runtime_enable(dev); > > + regcache_cache_only(priv->regmap, true); > > How does the cache get turned off again when runtime PM is configured > off? I'd expect the device to probe powered up and then be powered down > and put into cache only mode as a result of runtime PM running.
Thanks for the comments. Yes. need to consider when runtime PM off case. I will refine it. Best regards Shengjiu Wang
